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 2010-05-11

00:04:09 Quit captainewkl (Quit: Page closed)
00:10:59 Quit ender` (Quit: Science can destroy religion by ignoring it as well as by disproving its tenets. No one ever demonstrated, so far as I am aware, the non-existence of Zeus or Thor - but they have few followers now. -- Arthur C. Clarke)
00:14:23 Join Zarggg_ [0] (
00:14:30 Quit Zarggg (Read error: Connection reset by peer)
00:14:41 Quit Zarggg_ (Read error: Connection reset by peer)
00:14:52 Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
00:15:00 Quit joeyg (Quit: leaving work)
00:15:04 Join Zarggg [0] (
00:20:59preglowamiconn: not, nothing immediate
00:21:23*preglow tired, nightie
00:26:19 Quit Llorean (Quit: Leaving.)
00:34:59mtis it possible to disable plugins building ?
00:35:52 Join apoelstra [0] (
00:35:54 Nick apoelstra is now known as joeyg (
00:36:13gevaertsHave a look at ENABLEDPLUGINS in the generated makefile
00:36:28gevaertsI'm pretty sure that changing that will help
00:39:25mtthanks :)
00:44:44 Quit wodz (Quit: Leaving)
00:46:25 Quit jgarvey (Quit: Leaving)
00:48:42 Quit efyx (Remote host closed the connection)
00:52:58 Quit bluebro ()
00:59:09 Quit Ipsi (Ping timeout: 240 seconds)
00:59:47 Join Ipsi [0] (
01:03:02 Quit solexx (Ping timeout: 276 seconds)
01:03:44 Join TechnoKat [0] (
01:04:14 Join solexx [0] (
01:08:04 Quit Ipsi (Ping timeout: 240 seconds)
01:10:52TechnoKatis there a random album playmode?
01:17:34 Quit Zarggg (Quit: Leaving)
01:21:19stripwaxTechnoKat - can you define that more precisely? Do you mean something like 'shuffle' or 'random album change when finishing this album'?
01:21:33TechnoKatat the end of each album
01:21:37TechnoKatswitch to a new random album
01:21:41TechnoKator folder really
01:27:45 Quit flydutch (Quit: /* empty */)
01:31:44mc2739TechnoKat: there is a setting under playback settings for Auto-change directory. One of the options is random. This requires a configuration file which can be created by running the random folder advance config plugin. This is all described in the manual.
01:32:05TechnoKatalright, i looked at that plugin
01:32:12TechnoKati'll look at it some more
01:32:42mc2739check the manual sections for playback settings and plugins
01:40:10 Quit halmi (Read error: Connection reset by peer)
01:40:54 Quit stripwax (Quit:
01:42:04 Quit TechnoKat (Ping timeout: 240 seconds)
01:46:19 Quit nima (Ping timeout: 246 seconds)
01:48:01 Join nima [0] (
01:48:28***Saving seen data "./dancer.seen"
01:49:51 Part toffe82
01:52:06 Join Strife89 [0] (
01:52:42 Join mikroflops [0] (
01:56:31 Quit mikroflops_ (Ping timeout: 246 seconds)
02:00:45 Join zouhair [0] (
02:16:07 Join Zarggg [0] (
02:24:44 Join Rob2222 [0] (
02:27:16 Quit Rob2223 (Ping timeout: 246 seconds)
02:29:34kisakrbutil + sansafuzev2
02:30:39kisakI told it to update the bootloader to get that +1hr runtime goodness
02:31:07kisakand it dropped a file called fuzepa.bin onto the mp3 player
02:31:55kisakI believe the fuzev2 wants fuzpa.bin when updating firmware
02:32:44kisakseeing as fuzpa.bin envokes firmware update while fuzepa.bin does not
02:39:48 Quit Zarggg (Quit: Zarggg)
02:50:19 Join S_a_i_n_t [0] (S_a_i_n_t@
02:53:20 Quit evilnick (Quit: Page closed)
02:57:03 Join Darkknight512 [0] (
03:04:38bieberI'm going to refactor my theme editor parsing code to work with parameter lists instead of coding individual parse functions for each tags (don't know why this didn't occur to me :/). For long term maintainability, would it be preferable to have configurations hard-coded as constants, or loaded from a configuration file?
03:07:24 Join ischeriad [0] (
03:09:15S_a_i_n_tJdGordon: Sorry for pinging you last night then leaving you hanging, I fell asleep for once (I know right?). I just thought I'd mention that my legacy items from the .sbs on shutdown have popped back up again. I thought you'd fixed them accidentally when you fixed the".sbs flashing on database init" but it must have been a fluke. I'm now seeing my battery and disk access icons drawn onscreen during (and after, which is the problem) shutdown. They fade gr
03:10:00 Join CGL [0] (~CGL@
03:10:38 Quit ischeria1 (Ping timeout: 240 seconds)
03:11:22S_a_i_n_tI can pastebin the .sbs if you'd like, I can't see anything wrong with it but perhaps you can enlighten me. I'm not sure what I could be getting wrong to cause this though, and it's not consistent. It only seems to happen approximately 3 out of 5 shutdowns.
03:12:06S_a_i_n_tI can increase the odds of it happening by turning the player on, then turning it off immediately.
03:12:52 Quit MethoS- (Remote host closed the connection)
03:14:43S_a_i_n_tbieber: I'm not really a coder, but it seems to me that it would be easier to maintain/update a theme editor if the tags were in a config file as opposed to being hard-coded.
03:16:36bieberIt should be easier, I guess my only real concern is whether it's a good idea to base critical functionality on a config file that has to be bundled with the program
03:17:10S_a_i_n_tI see your point.
03:24:52 Join kugel [0] (~kugel@rockbox/developer/kugel)
03:25:27 Quit kugel (Remote host closed the connection)
03:26:28 Quit Darkknight512 (Quit: ChatZilla 0.9.86 [Firefox 3.5.9/20100315083431])
03:41:02JdGordonS_a_i_n_t: zip up the theme and ill have a look
03:42:06 Join anewuser [0] (anewuser@unaffiliated/anewuser)
03:48:32***Saving seen data "./dancer.seen"
03:51:05 Quit adnyxo (Ping timeout: 268 seconds)
03:54:34 Join CaptainKewl [0] (
03:54:37 Join jl-k [0] (
03:55:03 Quit jl-k (Client Quit)
03:58:18 Quit CaptainKewl (Remote host closed the connection)
04:03:27 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:03:28 Quit pixelma (Disconnected by services)
04:03:32 Quit amiconn (Disconnected by services)
04:03:34 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:03:47 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:03:56 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:05:36 Join Blue_Dude [0] (~chatzilla@rockbox/developer/Blue-Dude)
04:06:12 Join jl-k [0] (
04:07:06Blue_Dudefml: Good point re: hotkey tags. I'll go back and revisit them. They'll probably be gone in the morning.
04:13:58zouhairHi, anyone here kind enough to take a look at my plight :
04:20:56JdGordonbieber: hey, what do you mean parameterised lists?
04:21:33bieberJust a list of the parameters to a tag and their types and properties
04:22:10 Quit CGL (Ping timeout: 246 seconds)
04:22:30bieberi.e. something like 'V':width(int),height(int) ... and etc. I have no idea what format I'll actually end up using, but that sort of thin
04:23:04bieberThat would be in a config file, anyways. If it's hard-coded, there'd just be a class that specified the parameters for a tag
04:23:28 Join Forsaken [0] (~chatzilla@
04:24:41JdGordonI had an idea to simplify the parser so it could be more easily used directly by the theme editor
04:25:00JdGordonright now each tag does its own parsing which means lots of extra work for you...
04:25:38bieberThat's why I wanted to replace that with a generic function that could just parse according to parameter lists
04:25:57JdGordonwhat would be nice is if the handler for each tag accepted a char** of parameters, so no parsing was done in them, the same parser could be used by both
04:26:15JdGordoni.e the parser would split %V|a|b|c| into a usable array by both
04:26:59bieberSounds like a good idea, thanks
04:27:33JdGordonthe problem is that to do that the format has to change, which was the talk about %xx(a|b|c)
04:28:51JdGordonhow much work have you done on your parser?
04:29:29bieberI could still write a simple parser to recognize %ident(param'|')* or something like that
04:30:05bieberThe parser I have now is fully functional for the tags I've implemented, and for the rest it's going to be mostly boilerplate, as they're generally pretty easy to parse
04:30:16bieberBut that's still a lot of boilerplate, which is why I want to generalize the process
04:35:55JdGordonhmm... actually we could probably go directly to the full token array without using any special handlers? just copy the params into buffers and then handle them all at the end?
04:37:05bieberI do like the ability to use polymorphism instead of massive switch statements, but that could be preferable to having a separate class for all those tags
04:37:38JdGordonthis could also mean much better debug abilities
04:39:08JdGordonwell, now if a tag fails to parse itself it just fails and says which line it was on.. with this we could say "token xx expected a int for arg 3 but got 'a'"
04:39:43bieberThat would be more helpful
04:41:16JdGordonok, assuming tags must be in the form %aa %aa(b) or %aa(b|c|...) could a simple text->token array be done in c that we could share easily?
04:41:56bieberVery easily
04:41:59JdGordonthe bloody ; "tag" is going to be annoying to handle
04:42:37bieberThat would actually be regular, if all the parameters were guaranteed to be
04:43:11*JdGordon forgot conditionals
04:45:23bieberAside from the conditional tags, do the others all require simple values as their parameters?
04:46:25JdGordonall except the playlist viewer which right now does its own tag parsing on the last 2 params
04:47:02JdGordonbut I guess for that it could just run the full parser on them again
04:47:27JdGordon%Pv|0|%in - %ia|%fn|
04:48:04bieberYep, it'll be the same way for the conditionals: I'll just recursively call the document parsing function on them
04:49:18JdGordonif an array isnt good for the final list we can fiddle with the displayed to work more nicely with whatever
04:50:08bieberYeah, it'll have to be some kind of tree structure
04:50:35bieberAre you thinking about something that could be integrated into the theme engine or something? Since you were talking about a plain char* array and C code
04:51:18JdGordonyeah, I'd want to use the exact same code in rockbox for as much of the parser as possible
04:52:07JdGordononce the text is split into tokens and params you can go all c++y :)
04:52:28bieberThat sounds like a good idea
04:54:07bieberIf we're going to go down that road, I'd like to take it a step further and write a verifier the same way that could take the raw data and turn it into a parse tree that identified all the tags with enum values and checked their parameters and such
04:54:48bieberIf the theme engine people integrated that code into the theme engine, then the editor could just link to the files there, and at least parsing compatibility would be automatic, if not necessarily display
04:55:02JdGordonsounds good
04:55:20bieberDo you work on the theme engine? Or do you know if anyone would be interested in pursuing that?
04:55:26JdGordonright now checkwps does the ferification, but as i said, the debug output sucks
04:56:15 Join dys` [0] (
04:56:20 Quit dys (Ping timeout: 276 seconds)
04:57:49 Join Barahir_ [0] (
04:58:30 Quit jfc (Read error: Connection reset by peer)
04:58:37JdGordonthe lib could be done in a way that it doesnt even care about known tags at all, just assume anything after % is always a tag, keept the identifier in plaintext then at the end turn the returned parse tree into a usable token tree/array/whatever
04:58:53 Join jfc [0] (
04:59:11JdGordonor does that sound slow?
04:59:25bieberI don't think speed is going to be an issue, really
05:00:05bieberThat part is easy, I just figure that if I'm going to do the syntax parsing C-style, I might as well do verification too, so they can both be handled by Rockbox and the Theme Editor from the same code files
05:01:14JdGordonthe verification needs to know what each tag requires for its params, so wouldnt that be seperate?
05:01:22 Quit Barahir (Ping timeout: 258 seconds)
05:01:44 Quit soap (Ping timeout: 240 seconds)
05:03:12S_a_i_n_tJdGordon: <−− Here's that .zip you asked for.
05:03:26JdGordonso anyway, if you can do the parser, I can make the changes so the new tags will actually work untill your parser can be integrated?
05:03:29 Quit Galois (Ping timeout: 276 seconds)
05:03:44JdGordonS_a_i_n_t: nano right?
05:04:05 Join Topy [0] (
05:04:10 Quit Topy44 (Ping timeout: 248 seconds)
05:04:44 Quit bieber (Quit: No Ping reply in 180 seconds.)
05:04:59 Join bieber [0] (
05:05:37S_a_i_n_tJdGordon: It's the battery and disk access icons that get left onscreen after the player is doesn't happen consistently, pnly about 3 of 5 times.
05:05:39 Join jpt9 [0] (~jpt9@unaffiliated/jpt9)
05:05:47S_a_i_n_t*only too
05:05:53jpt9Hey. How well does Rockbox work on Sansa Clip v2s?
05:06:03jpt9(I have one, and I'm actually pretty happy with the OF...)
05:06:46jpt9(I know it's considered unsupported on the v2s...)
05:10:20JdGordonbieber: did you see my reply?
05:10:39bieberI don't think so
05:10:46bieberI got nothing after the last thing I sent
05:10:53 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
05:11:45JdGordonthe verification needs to know what each tag requires for its params, so wouldnt that be seperate?
05:12:30*JdGordon sent a email to the dev-ml about the format changes
05:12:47LloreanWhy exactly is the format changing?
05:13:09JdGordonto make it easier to read and parse
05:13:28bieberIt may or may not be, I think it's probably best to keep them separate
05:13:49bieberI could still do it in the same style, though, and both files could be used in the theme engine if desired
05:13:52LloreanShouldn't there be a significant discussion on what is "easier to read" then? Maybe hammer out a specific decision at devcon for how the tags should look, when a group can sit down and discuss it?
05:14:15 Join shai_ [0] (~Shai@
05:14:33LloreanI mean, your easier to read might be someone else's spaghetti, so one or two people changing the tag format (in a way that breaks compatibility, at least) probably shouldn't happen
05:14:40 Join soap [0] (~soap@rockbox/staff/soap)
05:14:53JdGordonbieber: unless you just mean if the syntax is correct.. otherwise you have to know what each tag exepcts to do proper verification
05:15:56 Join Orion777 [0] (
05:16:28Orion777does sansa fuze v2 require using .31 OF for rockbox?
05:16:30LloreanI mean, for example, I don't understand how () is any different from <> for lists.
05:17:10JdGordon<> means conditional
05:17:26bieberJdGordon: I mean verifying that it's semantically correct. They would still be separate functions, but you could chain them together and end up with a verified parse tree at the end
05:17:37JdGordonsure, ok
05:17:45 Quit shai (Ping timeout: 265 seconds)
05:18:25bieberSyntaxTree* parseSyntax(FILE* fin);
05:18:25bieberParseTree* verifyTags(SyntaxTree* rawData);
05:18:25DBUGSent KICK bieber to server
05:18:25bieberor something like that
05:18:25Kick(#rockbox bieber :No flooding!) by logbot!
05:18:54 Join bieber [0] (
05:19:09LloreanJdGordon: So the idea is to make it quickly apparent when something is a conditional, and when it's something with parameters?
05:19:14JdGordonstruct syntaxTree *parseSyntax(char *text) would be nicer :)
05:19:28JdGordonLlorean: yes
05:19:40bieberYeah, I guess that will work better if you want to embed it in Rockbox
05:19:51LloreanMakes sense
05:20:08bieberAnd I apparently just got kicked for flooding? What's that mean, exactly? (IRC noob :/)
05:20:18JdGordontoo many lines too quickly
05:20:19LloreanSo you end up with a generic %Xd tag that takes which image to display as its parameter?
05:21:02bieberAh, got it
05:21:14LloreanJdGordon: There should *definitely* be a discussion of tag letter changes though (as per your possible idea in the email)
05:21:37LloreanThere are a lot of them that can be very confusing because they seem related to other tags when they're not.
05:21:54JdGordonfeel like putting a proposal forward?
05:22:36LloreanI might. I don't write themes, so I don't know if it might not just be better to leave the tags relatively close to what people have already learnt.
05:23:30LloreanBut if I were to reorganize it, I'd probably make the first letter always a category letter. For example, most %pX tags are "playlist" related, but then you have %pV
05:23:34 Quit Orion777 (Ping timeout: 260 seconds)
05:24:16JdGordonI'm thinking time to change to 3 letter tags even
05:24:28LloreanThat might improve things a lot.
05:24:38LloreanMore room for clarity
05:25:40JdGordonwe are at an annoying point in the release window now... so if we assume this doent go in till staright after the next release we have a few weeks to make that perfect
05:26:06JdGordonexcept I think the other changes should be quick so bieber's new awesome tokeniser can be usable soon :)
05:27:04LloreanIt'd be nice not to break themes until after the release. What would you use the new tokeniser for in the release version?
05:27:15LloreanI imagine anything else that would benefit from it would also be big enough that it'd probably be a good idea to wait.
05:27:56JdGordonnothing, but the theme editor (or at least the parser) should work on known working themes
05:28:01bieberWhen you talk about three letter tags, would they all be three letters, or is that optional?
05:28:42JdGordonright now tags are 1 or 2 letters, usually 2. I'd say up to 4 letters with most being 3
05:29:11LloreanJdGordon: I don't know if that's a good reason to have experimental new code in the release.
05:29:27JdGordonyeah, probably best to branch
05:29:40bieberWould it be the same structure you have now, where the first letter classifies it as part of a group of similar tags?
05:29:41JdGordonjust thinking we need to work with GSoC timelines also
05:29:48LloreanAren't there supposed to be SVN branches for the GSOC students to work in?
05:30:26Lloreanbieber: At the moment, the first letter doesn't classify anything. There are some patterns, but it doesn't actually mean anything but a word. So "p" can mean "playlist" or "percent" or anything else that starts with P.
05:30:56bieberOh, okay
05:31:07LloreanIt'd be really, really nice if there were some classifications though
05:31:29LloreanIf we said "use p for playlist, m for metadata, a for audio state, f for file properties..."
05:31:33JdGordonespecially now that we have more than 1 screen where tags are usable
05:31:48LloreanAnd three letter tags would help keeping that flexible.
05:33:11JdGordonABCD - A being optional is the skin it works in (Wps, Base, Fm, Rec, etc), B being general group (Metadata, Settings, etc), C being more specific group, D being tag
05:33:51LloreanMaybe A not being optional, and having a code for "all screens"
05:34:37JdGordonthis part of the topic needs to be discussed, probably in the forum where the themers generally are more active
05:35:57LloreanSeems like a good idea.
05:42:23JdGordonactually i wonder if 4 is too many, making lines way too long
05:43:05bieberHow many tags do you typically have on a line
05:43:48LloreanSome people pile an awful lot of tags on a line.
05:43:58LloreanBut in many cases they probably just don't know they can split lines up
05:44:03JdGordon30+ isnt unheard of when using conditionals
05:44:21LloreanJdGordon: Though to be honest, a lot of those are volume bars and such, which can be replaced with progressive bars.
05:45:01bieberWhile we're changing things, why not allow multiple-line conditionals and the like without the comment hack?
05:45:55JdGordonyes, but how? require the line end with the |?
05:46:00 Quit Forsaken (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
05:46:41bieberThat could work
05:47:00LloreanI think the comment "hack" as it were looks / is pretty good actually
05:47:12LloreanComing at the end of the line, it's a unique line-ending character.
05:47:27bieberI guess that makes sense
05:47:47bieberYou could also consider the \ that a lot of formats use to allow explicit line continuations
05:48:19JdGordon%<line end> would be more consistent here
05:48:33***Saving seen data "./dancer.seen"
05:49:12LloreanI'd say \ or just #. A single character is nice, I think.
05:49:37LloreanThe # is pretty much going to have to stay anyway, unless comments change to not eat lines (or can't be done mid-line) in the future.
05:51:55 Join Strife89|Tablet [0] (
05:53:28 Quit jl-k (Quit: CGI:IRC (EOF))
06:02:49CIA-5New commit by jethead71 (r25940): MRobe100, HD200, Gigabeat F/X (anything with wm8750 or wm8751). Properly prescale bass and treble controls to avoid clipping.
06:13:06CIA-5New commit by jethead71 (r25941): r25940 oops. No, no, no, that wasn't quite right. Must be truthful about the level in decidecibels for subsequent code.
06:15:51 Nick shai_ is now known as shai (~Shai@
06:20:41 Join Orion7 [0] (
06:21:15Orion7anyone have experience with sansa fuze/clip
06:23:29Orion7I'm wondering if i decide to reset from OF .31 back .26 if it will effect my files(songs)
06:23:51Orion7and if i have to upgrade to .31 instead of .28
06:30:08 Join Galois [0] (
06:30:31 Quit Strife89 (Quit: Bed. ZZZZZzzzzzzzzzz......)
06:33:41HillshumOrion7: This is probably off topic.
06:34:09HillshumIt shouldn't, but you never know.
06:34:57Orion7ok, more specifically does rockbox on fuze v2 require .31 OF
06:41:17CIA-5New commit by Blue_Dude (r25942): More hotkey code cleanup
06:42:00 Quit anewuser (Ping timeout: 268 seconds)
06:42:01 Quit Strife89|Tablet (Quit: Leaving)
06:47:43 Join LinusN [0] (~linus@rockbox/developer/LinusN)
06:48:27 Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
06:55:22 Quit Orion7 (Ping timeout: 240 seconds)
07:08:31 Quit Horscht (Quit: Verlassend)
07:31:02 Quit grndslm (Quit: Leaving)
07:33:19 Nick dys` is now known as dys (
07:33:21amiconnjhMikeS: Shouldn't 1/10 decibel be called centibel, the same way as a decibel is in fact 1/10 bel?
07:33:36 Join komputes [0] (~komputes@ubuntu/member/komputes)
07:34:48 Join grndslm [0] (
07:48:37***Saving seen data "./dancer.seen"
07:51:27 Join Orion7 [0] (
07:51:57Orion7anyone around
07:54:36jhMikeSamiconn: I suppose it should. It was a bit smart arsy though. "tenth of a tenth of a..." :) any smaller and they're "tinkerbels"
08:03:40 Join efyx [0] (
08:05:01 Quit komputes (Ping timeout: 240 seconds)
08:12:28Orion7anyone here use a fuze v2
08:15:32S_a_i_n_tJdGordon: I take it you didn't find anything with that theme that would cause an unclean shutdown? AFAIK all the syntax is correct, I actually happen to think it's a pretty well written theme myself...
08:15:48S_a_i_n_tGimme a yell if you find something though
08:28:08 Join ender` [0] (
08:30:45 Quit tchan (Quit: WeeChat 0.3.3-dev)
08:34:25 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
08:46:51Orion7ender,tchan are you there?
08:47:14Orion7or anyone
08:48:53 Join robin0800 [0] (
08:57:51 Join cYmen [0] (
08:59:30 Join anewuser [0] (anewuser@unaffiliated/anewuser)
09:00:52 Join wodz [0] (~wodz@
09:01:54S_a_i_n_tOrion7: there is 118 people in the room, if sure if you ask it, someone can answer your question.
09:02:01S_a_i_n_t...or at least try to.
09:02:37 Quit antil33t ()
09:02:52Orion7well I have a fuze v2 question and I haven't gotten a reply yet
09:03:01 Join flydutch [0] (
09:03:43Orion7seems like noone while i've been here uses a fuze v2
09:04:37wodzOrion7: I can only advice a patience
09:05:26wodzOrion7: You can also look at IRC logs to see who is working on fuzev2 port
09:05:39 Join petur [0] (~petur@rockbox/developer/petur)
09:05:55Orion7on the rockbox site?
09:06:19S_a_i_n_tthere is a link to the logs in the topic
09:09:56Orion7hmm i'm not a coder
09:10:27pixelmaOrion7: I'd suggest asking your specific question - maybe it's not directly related to the Fuze or someone can answer it anyways without having this exact player
09:11:16wodzpixelma: how mem_test on M5 went?
09:11:31Orion7I want to know whether I have to use .31 OF and therefore will there be a volume cap at normal for rb
09:12:09 Quit markun (Ping timeout: 248 seconds)
09:12:37 Join jl-k [0] (
09:15:53 Quit sevard (Ping timeout: 248 seconds)
09:17:56pixelmawodz: got some numbers for you and I found out that the literal "boosted"/"unboosted" line doesn't mean anything. If I watch the clock displayed now it makes sense how read and write speeds change but the first line doesn't change with it (or just sometimes) and is at least half the time wrong
09:18:38 Quit jl-k (Quit: CGI:IRC (EOF))
09:19:44 Join markun [0] (~markun@rockbox/developer/markun)
09:19:50pixelmahere it goes: unboosted (clock) DRAM rd: 6.2 MB/s - wr: 9.4 MB/s, IRAM rd: 40.5 MB/s, wr: 40.0 MB/s
09:20:52pixelmaboosted DRAM rd: 17.3 MB/s - wr: 25.8 MB/s, IRAM rd: 110.3 MB/s - wr: 110.3 MB/s
09:21:06wodz*THAT* is the difference 200+ kB/s vs 6.2 MB/s
09:21:08pixelmanumbers sometimes change slightly
09:21:23 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
09:21:28 Join DerPapst [0] (~Alexander@
09:21:49wodzok I'll run new test and compare, thank You
09:22:58 Join B4gder [0] (~daniel@rockbox/developer/bagder)
09:25:18 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
09:26:35amiconnwodz: Your patch looks ok for the trailing word of unaligned reads (but two instructions can be saved). However, the leading word now has the opposite problem as before
09:27:01wodzamiconn: I just saw that and reworking this
09:27:18amiconnInstead of not being swapped when there was no word, it is now swapped twice when there is one
09:28:12amiconnThere are two possible solutions: either you put the lsr #8, r2 in an alternate code path (that means an extra branch), or you handle the leading 16 bit word the same way as the trailing one
09:28:23 Join komputes [0] (~komputes@ubuntu/member/komputes)
09:28:30wodzthis $#%#$# byteswap is a pain
09:28:47*amiconn agrees
09:29:50Orion7amiconn, do you know much about the fuze v2
09:29:59amiconnActually you might also use the trapf trick instead of a branch
09:32:55 Quit mc2739 (Ping timeout: 265 seconds)
09:33:21scorchemarkun: you should come to devcon
09:34:45 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
09:35:32JdGordonS_a_i_n_t: havnt looked yet
09:37:30cYmenWhat is the detailed status of rockbox on the sansa clip+?
09:39:05S_a_i_n_tNo worries JdGordon, I'm not entirely convinced the problem lies with the .wps/.sbs but I'm happy to be proved wrong. It will be a lot easier to track down the problem if the error is mine.
09:39:15pixelmaOrion7: for the third or fourth time - just ask your specific question. You are more likely to get help or an answer if you do so
09:39:30S_a_i_n_tI'm just confused about the erratic behaviour of these unclean shutdowns,,,
09:42:41S_a_i_n_tif it happeded every time I'd be more sertain about it being an error with my theme, I havent seen it happen with any of my other themes, though... I haven't seen using any other themes for a long time so that doesn't surprise me. I can't really say if it confined to this one theme or not.
09:43:29wodzamiconn: is trapf alias for tpf coldifire mnemonic?
09:44:45JdGordonS_a_i_n_t: umm... it probably depends where you are shutting down from
09:44:46S_a_i_n_tOrion7: If you are worried about there being a volume cap on RB, don't be.
09:44:51JdGordonfrom inside the wps or the list
09:45:15Orion7k ty saint
09:45:24S_a_i_n_tOrion7: a cap on the OF won't effect Rockbox at all.
09:45:36pixelmaOrion7: sorry, didn't see your question but S_a_i_n_t is correct
09:45:56Orion7np i'm new at this
09:46:34Orion7thanks again
09:47:30S_a_i_n_tJdGordon: It has happened before (icons left on the screen during/after shutdown) if it shuts down in the wps/file-tree, and it doesn;t seem to matter if it is an idle shutdown or a manual one.
09:47:53S_a_i_n_tIt seems to be more likely to happen if I turn the player off immediately after turning it on though.
09:48:40***Saving seen data "./dancer.seen"
09:48:55S_a_i_n_tI have had a really hard time trying to work out if it is random, or whether there is a recipie for it.
09:49:14JdGordonpixelma: I'm going to commit FM skin tonight, so feel free to replace the inbuilt skin when you are ready... I want to get it in and done before starting on the skin changes (longish reply to the dev ml being written)
09:55:08pixelmabtw. I mean the hardcoded FM screen in your patch in general if you haven't changed it since the versions I tried a while ago. They are different to what there is now and while I can understand the one change to get the progress bar in, I have a problem with putting the "Station:" lang string and the station name on one line because it will start scrolling earlier - especially on small screens
09:56:45JdGordonok sure
09:57:00JdGordonnothing says it cant be changed after its been commited
09:57:08 Quit Orion7 (Quit: State Farm is there.)
09:57:19JdGordonI just really really want to get it off my plate
10:00:12 Join swilde [0] (
10:00:56JdGordonscorche: can you add a subforum to themes for theme feedback?
10:03:18 Join kugel [0] (~kugel@rockbox/developer/kugel)
10:08:59 Join Orion7 [0] (
10:10:05Orion7for fuzev2 i'm trying to use mkamsboot on .31 OF but it always says only .26 OF works
10:12:14 Join Jaykay [0] (
10:12:31Orion7that only .17 and .26 are tested (Original firmware unknown, try a different version)
10:20:46wodzOrion7: guess why
10:21:25Orion7unfortunately I don't know
10:21:38Orion7because my understanding is I have to upgrade to .31
10:22:03GodEaterwhere have you got that idea from?
10:22:32Orion7I already have .26 running
10:22:39wodzOrion7: mkamsboot states clearly "only .17 and .26 are tested"
10:23:06wodzit doesn't mean .31 will not work it means it was not tested
10:23:07Orion7yes, but how can i upgrade my firmware to itself
10:23:24 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
10:23:30Orion7right, i tried using .31 and it says unrecognized firmware
10:23:42Orion7said that for .28 also
10:24:00pamauryJdGordon: did you manage to get the usb screen work without polling ?
10:24:11wodzwhat's wrong with "upgrading" to the same firmware version?
10:24:36JdGordonmkamsboot will refuse to work on an unknown OF, so even if it is tested it wont be officialy working untill its fixed
10:24:42JdGordonpamaury: not yet
10:25:12 Quit Kitar|st (Ping timeout: 276 seconds)
10:25:43pamauryJdGordon: really ? It doesn't work at all ?
10:25:49Orion7I didn't know i could ugprade .26 to .26
10:26:05JdGordonwell I didnt really put much effort into it :p
10:26:26wodzOrion7: "upgrade" usually means only write this #$@#% file to flash and reboot
10:26:58Orion7well i did that with .26 put everything at root and couldn't get anything to happen
10:27:20Orion7this was after it was succeful with mkams
10:27:22kugelare you in mtp or msc mode?
10:28:30kugelswitch to msc
10:29:00GodEaterI thought there was a huge risk of permanent brickage with the AMS sansas when doing a bootloader install ?
10:29:02amiconnwodz: yes. You can't use it as a mnemonic anyway if you want to use the trick
10:30:00wodzamiconn: honestly I don't know this instruction and I don't know what trick You are talking about
10:30:17wodzamiconn: I am beginer in coldfires :-/
10:30:32 Join Kitar|st [0] (~Kitar_st@
10:31:07amiconntrapf is essentially a trap that is never taken ("trap false"). It's purpose is that you can shadow one or two other instructions, by putting them into the extension word(s) of trapf.w resp. trapf.l
10:31:45amiconnThis saves cycles if you have an if/else construction with just one or two instructions in the else path
10:32:05amiconnAn unconditional branch is 3 cycles, while a trapf is single cycle
10:32:24 Join sevard [0] (sev@
10:33:23wodzamiconn: so imidiate argument to the trapf is in fact decoded instruction right?
10:34:42amiconnIt's not decoded if instruction flow "runs" into the trapf instruction, but you can branch "into" it with a conditional branch
10:35:05amiconnThis saves the unconditional branch for skipping the else path
10:36:59kugelindeed :)
10:37:15 Join b0hoon [0] (~quassel@
10:38:48b0hoonHi. I've released the bootloader for the vibe and it meets all the conditions of the stable port now. i guess.
10:39:34amiconnWell it's at least a little improvement. ARM has conditional execution for (almost) every instruction...
10:40:54CIA-5New commit by jethead71 (r25943): ARM DSP: Make things a little more pipeline friendly. Reduce nonvolatile register stacking where possible. Routines now handle odd sample counts ...
10:43:39wodzfirst test on optimised read/writes passed :-)
10:47:38wodzamiconn: I have to use .word 0x51fb instead of trapf.l ?
10:48:08amiconnYes, or .word 0x51fa instead of trapf.w
10:48:35wodzthat's *really* weird
10:48:45amiconnDepends on whether you want to shadow one instruction, or one instruction with an extension word itself or two instructions
10:49:23wodzand what branches with syntax like bne.s 1f mean?
10:59:20 Quit Orion7 (Ping timeout: 248 seconds)
10:59:51pixelmaJdGordon: btw. if looks of the FM screen change, you should update screenshots in the manual and I bet the "WPS" syntax appendix would need some rework. Make sure to at least file a bug report about it and give any info a non-coder manual write would need
11:00:07 Quit zouhair (Quit: Leaving)
11:03:46kugelb0hoon: so go ahead and make the change in the appropriate places :)
11:04:21b0hoonkugel: ok, thank you :)
11:04:55kugelb0hoon: nobody can prove you wrong anyway :P
11:06:38b0hoonkugel: everything should be ok, AFAIK
11:07:03wodzha test_disk write&verify passed 5 times now time to bench this
11:07:18b0hoonnothing is missing i think
11:08:27CIA-5New commit by jethead71 (r25944): ARM DSP: fore some reason I neglected dsp_downsample completely. Do a small reordering.
11:10:04wodzpixelma: Your new results from test_mem are in line with what I have on MPIO
11:15:07 Quit JdGordon (Changing host)
11:15:07 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
11:16:46 Quit BHSPitMonkey (Remote host closed the connection)
11:21:14JdGordonpetur: can you pm me your address? I dont tihnk it makes much sense for me to hold onto the usb tracer, so get it to you and devcon can work out where it lives
11:22:54JdGordonany perl ninjas got some spart time to whip up a script to do the skin syntax changes no exsisting .wps and .sbs files?
11:24:30kugeldoes it need to be perl? :(
11:24:58JdGordonanything excpet lisp!
11:26:34pamauryJdGordon: what do you want to do ?
11:26:38JdGordonits goign to suck when the theme site goes to 0 working themes with the current build
11:26:39 Nick mapi is now known as Guest951 (
11:27:05ZagorJdGordon: please don't commit before we have a working converter
11:27:59JdGordonpamaury: replace all the tags in the form %aa|a|b| to %aa(a,b) and a few from %aaV to %aa(V)
11:28:09JdGordonZagor: na, i wont
11:29:01JdGordonlets skip the june release and blame GSoC :)
11:29:21pamauryJdGordon: in all .sbs and .wps files ?
11:29:33pamauryI can try to hack a little script if you want
11:30:08JdGordoncool, it will be slightly more complicated than above though.
11:30:10pamauryExcept if someone has a super-perl-regex-or-something-knowledge and can do it in a minute
11:30:32CIA-5New commit by jethead71 (r25945): ARM DSP: Remove a couple comments that no longer apply.
11:30:38*Llorean was thinking we should try to time a release just before/around/at when GSOC ends so that we can immediately commit anything relevant after GSOC and have the maximum in-SVN time for new features before the next release.
11:30:39*pamaury wonders if "slightly more complicated" means recode de WPS parser to convert
11:30:55 Join antil33t [0] (
11:31:13JdGordonnot quite :)
11:31:25Zagorpamaury: regex is probably a good idea
11:31:38pamauryJdGordon: give me the except conversion scheme and I'll try, I have no WPS/SBS knowledge so be precise :)
11:31:56JdGordonZagor: it isnt regular.. and its not line by line
11:32:31pamauryI can do it in C if it appears to be infeasable with perl/bash/... ?
11:32:31Zagorhuh? it's text
11:33:05pamauryregex is not the solution to everything, regex can't code even though code is text ;)
11:33:47Zagorregex is the solution a large set of text transformation tasks
11:33:58Zagordoing it in C is just... messy
11:34:07JdGordonI'll pastebin the requirements
11:34:28Zagorbut since I don't have time to do it myself right now, I'm not going to be grumpy about it
11:34:29JdGordonI'd feel more comfortable doing this in c myself, but perl/python would be nicer
11:35:06 Quit Jaykay (Ping timeout: 240 seconds)
11:35:39 Quit anewuser (Quit: for SELL 2 by the price of 1 now!)
11:35:58pamauryok, I'll do my best to do it perl/python even though I do not code myself in either of this languages ^^
11:37:56JdGordonmaybe it would be simpler to not do a generic transform and work on the changing tags only (for which there are maybe a dozen)
11:38:06JdGordonjust thinking that getting it right inside conditionals is giong to be tricky
11:39:11Zagorhow could anything be simpler? B-)
11:39:52JdGordonwould that actually work?
11:40:01JdGordonthat wont get the messy ones inside conditionals
11:40:06pamaurydepends on what you want to achieve
11:40:07*Llorean doesn't know regex at all but thinks that might depend on %aa being present.
11:40:50ZagorJdGordon: I don't know enough details of the wps format to say
11:41:41Zagorsure, except no parenthesis around \1
11:42:35Zagorbut yes, I fully understand that this is rather messy for someone not comfortable with regex details.
11:43:24JdGordonIf that got 90% I reckon that would be good enough
11:43:31JdGordonexcept no it wont work
11:43:54JdGordon%?aa<%ab|blaa|fds> would get wrongly done
11:44:31pamauryWhat is the expected result with this ?
11:44:32JdGordononly do that regex is you arnt inside <>'s would get 99%
11:44:45Zagorpamaury: see the mailing list. JdGordon is going to change the wps syntax.
11:44:48LloreanIt might take longer, but wouldn't it be safer to just use an existing list of tags and check each one individually?
11:45:01JdGordonor that
11:45:09JdGordonlike i said, only maybe a dozen are actually changing
11:45:17LloreanIf we can parse existing WPSes without ambiguity, this script should be able to be 100% at least for reading and identifying
11:45:23LloreanThe only problem is writing things back in a reasonable manner.
11:45:45ZagorJdGordon: that regex won't match inside <> since it assumes a trailing |
11:46:12JdGordonit would get the first one though, wont it?
11:46:22pamauryThe best way seems to use the existing parser and output the transformed file, keeping enough information about spaces, newlines and comment
11:46:46ZagorJdGordon: no, \w only matches alphanumeric characters, so no < or %
11:46:59JdGordonpamaury: yeah, but hoping for a simple script so it can be run of the themes on the server
11:47:00 Quit dys (Remote host closed the connection)
11:47:24LloreanI'm not sure we should run it on the themes.
11:47:33LloreanEspecially if we make the change immediately after the release.
11:47:40ZagorJdGordon: sorry, wrong response. no it will ignore it since the string doesn't match the pattern. (no trailing |)
11:47:51LloreanInstead we should provide an webpage where authors can run it on their own themes and resubmit.
11:48:05LloreanMake it trivial for them to fix their theme, but let them choose not to (at least until the first release with the new parser comes out)
11:48:32JdGordon%?aa<%ab|blaa|fds> %ab|blaa| would match it though?
11:48:36JdGordonLlorean: yeah, thats good
11:48:41***Saving seen data "./dancer.seen"
11:49:13ZagorJdGordon: no. still wrong pattern.
11:49:28JdGordonbut that one is legal and needs to be changed
11:49:32JdGordonone or more params
11:49:34LloreanJdGordon: Couldn't you just run your test cases through the regex and see if they come out wrong. :)
11:51:09ZagorJdGordon: you can do variable params, but the regex won't be any easier. perhaps it's simpler to just run one regex for each number of parameters
11:51:42JdGordonYUCK :p
11:51:58*JdGordon makes a list of what the definite changes are
11:52:23 Quit komputes (Ping timeout: 240 seconds)
11:52:46JdGordonalso, - is a legal char in args
11:53:00JdGordonand numbers, but that is hit by \w anyway?
11:54:34Zagor' \w Match a "word" character (alphanumeric plus "_")'
11:55:10Zagor is a good reference
11:56:43*kugel thinks a fixed list of to-be-converted tags, and converting each alone is going to be simpler and safer
11:59:27LloreanWe'll need an explicit list of converted tags anyway, so we can post somewhere to let authors know (so they don't have to dig through the revision history on the theme page to figure out what all changed)
11:59:51*JdGordon is working on that now
12:01:29JdGordon25 changes
12:02:56 Join mikroflops_ [0] (
12:04:57 Join animalstyle [0] (
12:05:40 Quit animalstyle (Remote host closed the connection)
12:06:47 Quit mikroflops (Ping timeout: 240 seconds)
12:07:27*pamaury remotely extract JdGordon's list from his head and put it on pastebin
12:08:59 Join watto [0] (~watto@
12:12:34JdGordonalmost done
12:12:41 Join komputes [0] (~komputes@ubuntu/member/komputes)
12:15:00JdGordonmost of it is obvious, just replaceing |'s with ()'s and commas
12:15:59JdGordonotherwise its putting ()'s around the extra bits of tags that dont use |'s
12:17:09S_a_i_n_tJdGordon: Well, just make sure to put up a clear list somewhere of what's changes, with example syntax, so I can update my themes :)
12:17:38JdGordonthat's it for the first bit
12:19:41S_a_i_n_tso it's like " %Xx[x] " now? and like " %?Xx[blah,blah,blah] "?
12:19:50S_a_i_n_tI'm not quite sure I get it yet...
12:19:51pamauryJdGordon: when so you say ",[optional number])" you mean you have to put the comma in any case ?
12:20:14JdGordonno, that must be a typo
12:20:22*pamaury can't write today
12:21:12JdGordondecimal means a number with a possible .
12:21:17 Join VCR-clock [0] (~wut@
12:22:24CIA-5New commit by jethead71 (r25946): ARM DSP: Get an instruction out of the crossfeed main loop and improve loads and interlocks for stacked vars.
12:28:40kugelJdGordon: why does %Vi has the label optional but for %V it's a separate tag %Vl?
12:28:59JdGordonjust the way I happened to write it
12:29:14JdGordontrying to make it easier for whoever does the script
12:29:27 Quit VCR-clock (Quit: Unplugged.)
12:29:49S_a_i_n_t <−− New 16px Tango! iconset (+simple viewers), on DAP testing with "large screen" type players would be appreciated, you'll want to use at least a 16pt font as well obviously. Enjoy ;)
12:32:48kugelJdGordon: I just found that inconsistent
12:33:10kugelwe could remove %Vl, couldn't we?
12:33:14JdGordonoh right
12:33:18JdGordonI see what you're saying
12:33:25JdGordonyeah, we could
12:33:42*kugel would rather have optional params at the end
12:34:15JdGordonlets do this break first then figure out which tokens really need replacing/fixing
12:34:56kugelbreak another time just after the big breakage?
12:35:48JdGordonas long as it happens in one release cycle its fine
12:36:04JdGordonor we really want to completly destory everything?
12:36:12JdGordonin which case there isnt much point having a convertor is there?
12:36:19kugelthe first breaks it completely already
12:36:30JdGordonrelativly simply though
12:36:37kugelremoving %Vl could be part of the converter
12:37:13JdGordondo i just do svn branch to create one?
12:37:38kugelno, you do svn cp
12:38:20kugeldo you agree with optinal parameters at the end? that helps consistency
12:38:29*amiconn usually uses as a reference
12:38:56JdGordonthe placement doesnt really bother me
12:39:10kugelI think "%xd<one or two letters) -> %xd(letters)" should be "%xd<one or two letters) -> %xd(letter 1,letter 2)"
12:39:11JdGordonI would like general concsistancy across all tags though
12:39:37kugelwhich is exactly what I'm saying
12:39:40JdGordoni.e the id field should be the first for all things that have a id, optional or otherwise
12:40:15JdGordonand no, if we do your idea than %xd(a,b) would be different to %xd(ab)
12:40:32JdGordonthe first being enable the a and b images, the second is enable the 2nd subimage in a image
12:41:37kugelJdGordon: not if we do the other idea of different delimiters for repeater params
12:42:10kugelI think it should be clear that one is a parameter list and the other is just a simplification for repeated tags
12:42:43JdGordonan example?
12:43:21kugel%xd(A,a) <-> %xd(A;B)
12:43:43kugelcould even do %xd(A,a;B,b) then
12:44:00 Join Forsaken [0] (~chatzilla@
12:44:17JdGordonthat doesnt look very nice
12:44:28kugelif there's a ; (assuming that's the repeater delimiter), it would just call the tag parser multiple times for the args
12:45:38JdGordonapart from the xd tag, where else would that be useful?
12:46:03JdGordonit would be simple enough to just put that logic stright into those tags
12:46:30kugelwhy don't make it generic?
12:47:06JdGordonit feels messy, but either way, adding that wont break anything
12:47:26kugelI don't think the repeater thing is very important, we can worry about it later as adding it later shouldn't break things
12:47:41JdGordonso the origional point is that %xd(aa) is simpler than %xd(a,a)
12:47:50JdGordonnot simpler... umm... flexible
12:48:09kugelit's only inconsistent imo
12:48:24JdGordonnot really.. aa is an id
12:48:59kugelif you follow %xl, then the subimage is an additional, optional parameter
12:49:02kugelso it should be for %xd
12:49:31kugeldoing %xl(A,a) but %xd(Aa) is incosistent
12:51:31JdGordondoing that means we have no choice but to use a different delimiter (which will make the parser slightly omre complex) or being stuck with long lines of %xd(...)'s
12:51:59kugelI'm very much in favor of a different delimiter
12:52:05JdGordonor.. %xd(A,d,B,F,d,e) is slightly ambiguous
12:52:14kugelusing the same for parameters and repeaters is confusing
12:52:17JdGordonof course we could know that B doesnt have subimages
12:53:46kugelwe could with unecessary extra logic in the parser
12:54:03JdGordonputting the repeater logic in the handler will dupe some code but make the important parser simpler
12:54:07kugela separate delimiter is easier for parsing and less confusing for themers
12:54:18JdGordonmaking it generic for 2 or 3 cases doesnt really make much sense
12:54:35kugelgeneric is less code though
12:54:53JdGordonin a more complex area
12:55:58wodzwhy don't we use xml-ish format?
12:56:09kugelI think the parameter parser would be the least complex part, the individual tags are more complex
12:56:38*JdGordon glares at wodz
12:56:41JdGordonxml is never the answer!
12:57:01wodzbut is easy to parse
12:58:11JdGordonkugel: my tihnking is this, I want the parser to split up the args and pass them as a char** to the handler which then works on the individual char*'s.. the debbuger could then very easily know if it got a number when it wanted a filename
12:58:23JdGordonthe tag parsers then become very dump
12:59:35JdGordoni.e "did I get 5 params? good.. are they int, int, string, string, string? no? ok fail"
13:02:00 Quit kugel (Ping timeout: 240 seconds)
13:02:37gevaertswodz: is it?
13:03:44*Llorean still likes "one line of the WPS file equates to one line of the WPS"
13:03:58LloreanWith simple WPSes / text based WPSes it makes creating them and visualizing them very easily
13:04:01JdGordonthat isnt changing
13:04:11LloreanJdGordon: That was in response to the XML thing
13:04:41CIA-5New commit by jdgordon (r25947): make a new branch for the breaking changes for the skin engine
13:04:49*JdGordon hopes he dun that right
13:05:48*S_a_i_n_t wonders if he has to update his themes now?
13:05:59JdGordonnot yet
13:06:06gevaertsS_a_i_n_t: you could wait until there's actual code
13:06:09JdGordonunless you want to help test
13:06:21wodzIt was just a loose question. I am big XML fan but it has some strong points
13:06:36JdGordonI've done the code already for the existing code to work with the changes to the syntax
13:06:46wodzs/ big/ not so big/
13:07:10gevaertswodz: is availability of parsers that don't use more than a few KB one of those strong points? ;)
13:07:48wodzprobably not
13:08:53 Quit komputes (Ping timeout: 246 seconds)
13:09:39LloreanProposal for 3.6: Finally remove "Fade on stop/pause" as defaulting to on.
13:11:09Zagoroh right, is it still that silly long fade?
13:11:13LloreanZagor: Yes.
13:11:36Zagorbleh, that's a bug. we should simple fix the default to be as it was intended. (a replication of winamp's fade on pause)
13:11:50LloreanIt looks unresponsive because it seems to take a while to pause (it can look buggy) and (to a lot of the people we talked with about this) it's basically just a feature that immediately gets disabled anyway.
13:12:07LloreanI'm a big fan of most features disabling to "off" and needing to be turned on anyway
13:12:21Zagorit should be a fraction of a second, just to take the edge off
13:12:39LloreanIf it were very quick, it probably wouldn't be as annoying, yes.
13:13:10JdGordonif it were done right and not faked with fiddling with volume it would be less annoying
13:13:49Zagorit was done in hwcodec times
13:15:11LloreanWell, maybe it should be disabled by default for now, then we can ask people what they prefer when a properly working one exists?
13:15:50Zagorif it isn't fixed before the next release, I agree default should be "off"
13:17:07 Join lpereira [0] (
13:17:17S_a_i_n_tI don't think it should be off...but it should be a *very* short fade,
13:17:30ZagorS_a_i_n_t: right, that's what I mean with "fixed"
13:17:45S_a_i_n_toh, *facepalm*
13:19:56 Quit B4gder (Quit: It is time to say moo)
13:20:54 Join Blue_Dude [0] (~chatzilla@rockbox/developer/Blue-Dude)
13:22:10Blue_DudeSaw something about "fade on stop/pause". That's something I'm working on doing in the mixer. So at least swcodec will be fixed eventually. Hwcodec will still probably need to do it as a volume fade.
13:23:13S_a_i_n_tit just needs to fade out quite a bit quicker on pause than it does now...
13:23:28Blue_DudeIs 23 ms short/long enough for the job? Any multiple of that will work well.
13:23:48Blue_DudeIt will happen within 23 ms. :)
13:23:51LloreanI think the shorter the better.
13:24:09Blue_DudeI meant the length of the fade though.
13:24:14LloreanBut 23 might be a little too short to be noticieable.
13:24:18S_a_i_n_tthat would probably be nicer than it is now.
13:24:34S_a_i_n_tperhaps 46ms?
13:24:36LloreanI'm guessing (based off nothing at the moment) we'll probably settle on something between 100 and 200
13:25:17Blue_DudeIt will occur within 23 ms of the button press, but it can be any multiple of that for the fade itself. Keeps the code cleaner. (23 ms is the mix buffer size).
13:25:23LloreanIt may make sense to have a setting for the fade duration anyway, at which point we can all try it and see which seems a good default
13:25:42Blue_DudeBleh. I thought you didn't like menu items like that.
13:25:53LloreanThere's already a menu item to turn it off/on
13:25:59LloreanReplace it with "Duration" that starts at 0.
13:26:10LloreanNo net increase in menu items, just an increase in functionality.
13:27:46 Quit Forsaken (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
13:31:20Blue_DudeHm. What, 0, 23, 46, etc.? Not pretty.
13:31:41Blue_Dude"Off, Short, Long" maybe.
13:31:47 Quit lpereira (Quit: Leaving.)
13:32:01Blue_DudeWith hardcoded presets.
13:32:03LloreanOr "short, medium, long" being 46, 115, and 230 or something
13:32:14LloreanBut yeah, exposing the numbers would be ugly
13:32:27Blue_DudeSubject to experimental changes later.
13:32:49LloreanBut I think that if the feature is to exist, it needs a little flexibility otherwise it's really there for a very, very small subset of users.
13:32:56LloreanEither that, or it needs to be something that more or less universally sounds good.
13:33:05LloreanIn which case, it may not even need an on/off
13:33:34S_a_i_n_tI think even 46 would be barely audible, perhaps a slightly longer setting for "short" than 46ms
13:34:21ZagorS_a_i_n_t: yes. an enterprising soul might measure how long winamp fades.
13:34:45Blue_Dude"Barely audible" seems about right to me. I use a hard stop myself. Fade bugs me.
13:34:54S_a_i_n_talas, I do not have a winamp to measure.
13:35:03ZagorBlue_Dude: then 0 is for you
13:35:14Zagordon't ruin the function because you want it off
13:35:30S_a_i_n_tI like the fade on stop/pause setting, but, the fade is slightly too long IMO
13:35:33LloreanI think "barely audible" is good for short
13:35:43LloreanIt should take the edge off without being something you can notice unless you're looking for it hard.
13:35:43S_a_i_n_tit needs to be about halved in duration.
13:35:50Zagorthe feature as-intended is very nice. it's just the current incarnation that's silly.
13:35:52Llorean"medium" should be the setting for most similar to winamp
13:36:01CIA-5New commit by wodz (r25948): HD200 - fix in asm optimised ATA reads&writes
13:36:03LloreanAnd "long" should feel like it's actually fading.
13:36:27LloreanS_a_i_n_t: I'm pretty sure the current setting halved would still be well, well over the 230ms I mentioned for long
13:37:00Zagorthere should be no noticeable delay in pausing. just a smoother cut-off.
13:37:31Blue_DudeZagor: I'm not going to ruin it. :)
13:37:38 Quit wodz (Quit: Leaving)
13:37:50S_a_i_n_twe'll see.... ;)
13:38:34Blue_DudeWell, I won't sabotage it anyway. Ruin is always a possible side-effect.
13:39:27S_a_i_n_theh, good work with hotkey btw...but I have a request.
13:39:48S_a_i_n_tCould you please add "insert shuffled" to the file-tree hotkey?
13:40:08Blue_DudeCould be. What's it worth to you?
13:40:24Blue_DudeDon't answer that.
13:40:30S_a_i_n_tI kinda looked, but I couldn't really see how to do it myself.
13:40:40Blue_DudeYeah, that's a trivial addition. Consider it done.
13:40:58S_a_i_n_tthanks, I for one would find it very useful.
13:41:12S_a_i_n_tI use insert shuffled quite a lot.
13:41:15Blue_DudeBut I don't use shuffle so I might ruin it...
13:44:03 Join fleebailey33 [0] (~fleebaile@unaffiliated/fleebailey33)
13:47:23 Join DataGhost [0] (
13:47:24 Quit DataGhost (Changing host)
13:47:24 Join DataGhost [0] (~dataghost@unaffiliated/dataghost)
13:47:39 Quit amiconn (Read error: Operation timed out)
13:48:17 Join kugel [0] (~kugel@rockbox/developer/kugel)
13:48:43***Saving seen data "./dancer.seen"
13:48:54kugelI think I still have some code around that makes the fade asynchronous, i.e. the wps is left immediately
13:49:13 Join adnyxo [0] (
13:49:29 Quit pixelma (Read error: Operation timed out)
13:49:37 Part fleebailey33
13:50:14LloreanI think that would be an improvement independent of what is decided on regarding the fade time
13:50:32kugelyea agreed, it's too long too
13:56:20 Join dfkt [0] (~dfkt@unaffiliated/dfkt)
14:02:46 Quit pamaury (Quit: Quitte)
14:03:06kugelquick&dirty resync. I don't even know if it compiles. kugel-rb.git/shortlog/refs/heads/fade_on_pause">
14:20:16 Join amiconn_w [0] (~55b7c184@gateway/web/freenode/x-ziyrybdiggmhwfjh)
14:22:28 Join lpereira [0] (
14:24:04 Quit robin0800 (Remote host closed the connection)
14:29:28 Join komputes [0] (~komputes@ubuntu/member/komputes)
14:34:37 Quit lpereira (Ping timeout: 245 seconds)
14:34:59 Join lpereira [0] (
14:37:25 Join pixelma [0] (quassel@rockbox/staff/pixelma)
14:37:50CIA-5New commit by jethead71 (r25949): ARM DSP: Add assembly custom sound channel processing. 13% to 14% faster than currently-used default C code on ARMv4.
14:38:07 Join amiconn [0] (quassel@rockbox/developer/amiconn)
14:39:39 Part LinusN
14:43:27 Quit amiconn_w (Quit: Page closed)
14:54:27JdGordonlast call for testers for the fm skin patch...
14:58:57JdGordonkugel: %Xd to say the backdrop is disabled.. keep as a seperate tag? or %X(d) ?
14:59:07*JdGordon thinks the first
15:00:01kugel%X(d) would be ambigious
15:00:05pixelmait's a bit weird as %xd says "display the preloaded image" but it goes along with %we/%wd
15:00:28 Join lpereira1 [0] (
15:00:30 Quit lpereira (Read error: Connection reset by peer)
15:00:36JdGordoncapital x is backdrop
15:00:37kugelI just thought the same actually
15:01:05LloreanJdGordon: Could you do %X(-) to disable it maybe?
15:01:38JdGordonthat works also
15:01:48kugelI guess that would work, but - usually means "pick defaults"
15:02:06JdGordonI'm not too fused what it is, just want to work out if it should get ()'s or not
15:02:22JdGordonit cuold even be empty
15:02:47Lloreankugel: Isn't the true default "no backdrop"?
15:03:00LloreanI think it should get the parenthesis but either - or empty
15:03:16kugelthe default is the menu backdrop
15:04:03*kugel thinks plain %X makes most sense, leave it empty for an empty backdrop (ie no backdrop)
15:04:46 Quit komputes (Ping timeout: 240 seconds)
15:05:54Blue_DudeS_a_i_n_t: Just to make absolutely sure, you want "Insert Shuffled" right? Not "Insert Last Shuffled"?
15:12:15 Join JohannesSM64 [0] (
15:15:45 Join stoffel [0] (
15:18:27 Join komputes [0] (~komputes@ubuntu/member/komputes)
15:20:27 Join Schmogel [0] (
15:36:02 Join CGL [0] (~CGL@
15:36:58JdGordonoh joy! I've fixed the parser to handle (,,,)'s but now it segfaults the displayer! :<
15:39:09Blue_DudeI think that's got it...
15:39:22Blue_DudeIt was more of a challenge than I thought, but here you go...
15:40:27CIA-5New commit by Blue_Dude (r25950): Add Insert Shuffled to hotkey options
15:44:37 Nick fxb__ is now known as fxb (
15:45:13 Nick fxb is now known as fxb__ (
15:48:45***Saving seen data "./dancer.seen"
15:50:40CIA-5New commit by Blue_Dude (r25951): Fix yellow
15:51:26 Quit CGL (Ping timeout: 268 seconds)
15:55:09JdGordonparse_list() doesnt really like 2 seperators :/
15:56:50kugelparse them separately?
15:57:14 Join MethoS- [0] (~clemens@
15:57:30kugelfor each repeater delimiter do a parse_list() and parse_tag(). anyway, I think this can be done later so I wouldn't worry about it now
15:57:50JdGordonno... the problem is where it looks for the end of non-number items it looks for the seperator
15:58:00JdGordonbut now the last one is a ) while the seperator is ,
15:59:36JdGordonsimple enough to hardcode a workaround.. but not so nice
16:00:40 Part jpt9
16:01:20CIA-5New commit by jdgordon (r25952): initial work to move to (,) instead of | seperators.. ...
16:01:31 Join Blue_Dude_ [0] (~chatzilla@rockbox/developer/Blue-Dude)
16:01:35 Join pyro_maniac [0] (
16:01:39JdGordonalso %Sx which i forgot to add in the message :/
16:01:46 Quit Hillshum (Ping timeout: 246 seconds)
16:01:46 Quit joeyg (Quit: hwlp)
16:01:53kugelmaybe a max_len for parse_list() which could be gotten by strchr'ing the param list before?
16:02:53JdGordonfor now I just hardcoded ')' for the sep if the item is the last to be parsed
16:02:54kugelare you going to do the "char *params[]" idea?
16:03:46 Join CGL [0] (~CGL@
16:05:21 Quit Blue_Dude (Ping timeout: 265 seconds)
16:05:27 Nick Blue_Dude_ is now known as Blue_Dude (~chatzilla@rockbox/developer/Blue-Dude)
16:09:28CIA-5New commit by jethead71 (r25953): Gigabeat F/X: Let us clear up confusion about just what the core frequency is. Fix frequency display in buffering screen.
16:11:31 Join evilnick [0] (~0c140464@rockbox/staff/evilnick)
16:13:48CIA-5New commit by jdgordon (r25954): add %d() %Sx() %St() to the tested list. ...
16:19:01 Join apoelstra [0] (
16:19:06 Nick apoelstra is now known as joeyg (
16:30:29 Part b0hoon ("Back to work.")
16:30:36 Join Blue_Dude_ [0] (~chatzilla@rockbox/developer/Blue-Dude)
16:31:02 Quit kugel (Remote host closed the connection)
16:34:15 Quit Blue_Dude (Ping timeout: 248 seconds)
16:34:17 Nick Blue_Dude_ is now known as Blue_Dude (~chatzilla@rockbox/developer/Blue-Dude)
16:37:03 Quit Schmogel (Ping timeout: 268 seconds)
16:50:32 Join Schmogel [0] (
16:51:50 Join jgarvey [0] (
17:09:56 Part Zagor
17:09:57 Quit elinenbe (Read error: Connection reset by peer)
17:12:37 Join whydoubt [0] (
17:12:54 Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
17:14:02 Quit antil33t (Read error: Connection reset by peer)
17:14:09 Join antil33t [0] (
17:16:49 Join wombat23 [0] (
17:19:21 Quit jae (Quit: leaving)
17:20:40 Join jae [0] (
17:25:43 Quit MagusG (Ping timeout: 240 seconds)
17:32:52 Quit lpereira1 (Quit: Leaving.)
17:35:02 Quit dfkt (Ping timeout: 245 seconds)
17:42:36 Quit petur (Quit: *plop*)
17:48:46***Saving seen data "./dancer.seen"
17:51:09 Quit arun (Read error: Connection reset by peer)
17:52:56 Quit bluebrother (Disconnected by services)
17:52:59 Join bluebroth3r [0] (~dom@rockbox/developer/bluebrother)
17:53:02 Join Boldfilter [0] (
17:56:02 Join fml [0] (
17:56:28fmlWill the new WPS syntax allow for file names with a comma in them?
17:57:46 Join Strife89 [0] (
17:58:11 Join toffe82 [0] (~chatzilla@
17:58:26whydoubtI didn't see a tone generator plugin for rockbox so I wrote one. What would it take to get it into the repository?
17:58:46 Quit Schmogel (Ping timeout: 245 seconds)
18:01:18 Join dfkt [0] (dfkt@unaffiliated/dfkt)
18:02:40markunwhydoubt: put it in the tracker and get some people to use it/look at it and bug them to commit it :)
18:07:17Strife89What's this?
18:09:12 Part Strife89 ("Leaving")
18:09:15 Join Strife89 [0] (
18:09:44Strife89whydoubt: What targets have you written keymaps for? :)
18:10:57 Quit komputes (Ping timeout: 246 seconds)
18:14:40 Join Zarggg [0] (
18:14:46 Quit Zarggg (Read error: Connection reset by peer)
18:15:44 Join Zarggg [0] (~zarggg@2001:0:4137:9e76:0:fbf3:beb1:ba3d)
18:15:54whydoubtStrife89: I borrowed code from the metronome. PLA_INC/_DEC for volume, PLA_RIGHT/_LEFT for note, PLA_START to play/pause. Does that help or did I misinterpret your question?
18:16:17 Join n1s [0] (~n1s@rockbox/developer/n1s)
18:16:33Tornesimple pluginlib actions like that should work on just about every target, yes
18:16:43TornePLA only becomes problematic when you have to start combining different sets
18:16:45Strife89Ah, okay.
18:17:10Strife89whydoubt: Reason I was asking is, I can test it on four targets for ya. :)\
18:17:34 Join hebz0rl [0] (
18:21:17 Join Schmogel [0] (
18:21:36 Quit Schmogel (Read error: Connection reset by peer)
18:22:37 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
18:22:51evilnickDoes the Fuse sim (3.5 from rasher's site) show Internal memory and SD memory separately?
18:25:04 Join komputes [0] (~komputes@ubuntu/member/komputes)
18:27:32 Quit stoffel (Remote host closed the connection)
18:31:02 Quit xavieran (Ping timeout: 248 seconds)
18:33:00 Join stoffel [0] (
18:36:12kisakI want to fix a 1 character typo in rbutil.ini, is it better to make a diff and put it on flyspray or to get setup to commit to svn myself?
18:37:10 Quit grndslm (Ping timeout: 260 seconds)
18:38:47S_a_i_n_tkisak: *probably* best to add a patch to the tracker...
18:39:02kisakor should I just wait to give a shoutout to funman/domonoky/bluebrother
18:39:15topikor shout now. people read the logs
18:39:16*domonoky looks around
18:39:38 Quit Zarggg (Ping timeout: 260 seconds)
18:39:58*n1s ponders a good benchmark for strlen
18:40:21 Quit swilde (Quit: ERC Version 5.3 (IRC client for Emacs))
18:40:27domonokykisak: if you tell me the typo, i can also just fix it.
18:40:33kisakthe sansafuzev2 entry in rbutil.ini should have fuzpa.bin as the bootloader filename, not fuzepa.bin
18:41:43 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
18:42:21domonokyah, thats important, or else the upgrade wont work.. i will change it.
18:42:57kisakthank you domonoky
18:43:26CIA-5New commit by Domonoky (r25955): rbutil: fix fuzev2 firmware filename. Thanks to kisak for spotting it.
18:43:46 Quit komputes (Ping timeout: 245 seconds)
18:44:22 Quit Strife89 (Quit: Leaving Wendy's.)
18:44:27 Join xavieran [0] (
18:45:21kisakI noticed that yesterday evening when I went to bump the bootloader to take advantage of r25900/FS #11246
18:52:41 Join Zarggg [0] (
18:54:34 Quit jae (Quit: leaving)
18:56:20 Join komputes [0] (~komputes@ubuntu/member/komputes)
18:58:13 Join jae [0] (
18:58:35 Join wincent_balin [0] (
19:00:30 Quit wincent (Ping timeout: 260 seconds)
19:00:44 Join anewuser [0] (anewuser@unaffiliated/anewuser)
19:02:46 Quit pyro_maniac (Quit: Konversation terminated!)
19:05:32 Quit JohannesSM64 (Ping timeout: 246 seconds)
19:05:38 Join Jaykay [0] (
19:10:11CIA-5New commit by alle (r25956): No need to have a variable for a thing computed by the preprocessor
19:10:42 Quit S_a_i_n_t ()
19:18:01 Join S_a_i_n_t [0] (S_a_i_n_t@
19:18:05 Join JohannesSM64 [0] (
19:22:56 Quit fml (Ping timeout: 245 seconds)
19:27:56 Join stripwax [0] (
19:28:11 Quit stripwax (Client Quit)
19:28:26 Join PyroBor [0] (
19:30:40 Quit xavieran (Ping timeout: 240 seconds)
19:32:24 Quit Boldfilter (Quit: Boldfilter)
19:34:47PyroBorHellow. Found one mistake (couldn't call that a bug) in source code in file rbutil/INSTALL: "rbutilqt" should be "RockboxUtility". Just like it is mentioned in changelog on the website.
19:37:52PyroBoralso flyspray didn't like "č" in my name :)
19:39:09PyroBorthat mistake is in rockbox utility.
19:39:44*domonoky will fix..
19:41:38CIA-5New commit by Domonoky (r25957): rbutil: change the binary name here too. Thanks to PyroBor for finding that.
19:43:11 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
19:43:54 Join xavieran [0] (
19:44:30S_a_i_n_tBlue_Dude: Thanks a bunch for adding 'Insert Shuffled" to the file-tree hotkey settings! I appreciate it, very handy :D
19:46:04 Quit DerPapst (Quit: Leaving.)
19:48:48***Saving seen data "./dancer.seen"
19:53:13 Join kugel [0] (~kugel@rockbox/developer/kugel)
19:55:10 Quit n1s (Ping timeout: 240 seconds)
19:59:03 Part PyroBor ("Konversation terminated!")
19:59:58 Join liar [0] (
20:05:20 Join n1s [0] (~n1s@rockbox/developer/n1s)
20:05:47 Quit bzed (Read error: Operation timed out)
20:07:50 Quit ranma (Ping timeout: 260 seconds)
20:08:45 Join ranma [0] (
20:09:12 Join bzed [0] (
20:09:20*kugel wonders why buttons stop working correctly if I change the sdl threading a slight bit
20:11:57 Quit wincent_balin (Ping timeout: 240 seconds)
20:14:59 Join DerPapst [0] (
20:15:10 Join Buschel [0] (
20:18:35Buschelwodz: regarding your comparison of pixelma's test_mem results (several MB/s) with yours (some hundred KB/s) -> you need to use current svn's test_mem plugin. there were several changes in the last days to optimize for ARM and to correct the calculation of memory bandwidth. in your version you will have to multiply the result by 100 to have the real bandwidth (e.g. your 200+ KB/s equal 20+ MB/s).
20:20:59 Join wodz [0] (
20:21:23 Join moos [0] (moos@rockbox/staff/moos)
20:21:44wodzBuschel: I did compare current svn test_mem results and they are the same as pixelma's
20:22:09Buschelok. fine :)
20:22:55wodzBuschel: wouldn't it be better to use memset/memcpy in test_mem? coldfire has optimised versions thanks to amiconn
20:23:16wodzthe question is what we try to benchmark
20:24:00n1swodz: looking at your ata-as-mpio change, it looks like you could save an instruction in a few places by replacing and.l #0xff, %d0; or.l %d0, %d2 with move.b %d0, %d2
20:24:28wodzmove.b does not clear rest of register
20:24:41wodzso I can't do or next
20:24:43Buschelwodz: we could add this measure as well. but I like to have separate results for read and write. e.g. in codecs you may read large amounts of date from iram to save a single results in dram (like used in filters).
20:25:24Buschelwodz: memcpy or memmove would be more symmetric
20:25:32n1swodz: i mean replace the and and the or with the move.b just copying the lsB from d0 into the lsB of d2
20:26:02wodzn1s: aa right this will work
20:26:57wodzn1s: I wanted to review the code once more when merging into ata-as-coldfire.S
20:27:27preglowamiconn: planning to optimize the arm flac routines for higher bit depths?
20:27:35n1si definitely think a merge is a good idea, and one can never review too many times ;)
20:28:06wodzBuschel: but now we can't in any way compare results from different architectures
20:29:41Buschelwodz: as I said, we might add memmove and memcpy to have asm'ed versions for the different architectures. and in addition mr. someone might add asm for other architectures for the read/write tests :o)
20:29:52wodzBuschel: writing optimized routine for every architecture just for test_mem is waist of time
20:30:16Buschelthat is your opinion
20:30:43wodzwhat information does it gives you?
20:31:20Buschelimplementing it might be fun for some of us ;o)
20:31:22wodzit is much better to benchmark code which is used in some other places in rockbox
20:31:31 Nick YPSY is now known as Ypsy (
20:32:40Buschelwodz: you are missing some major point here: codecs (that's what I mostly work on) _are_ specific in their decoding/synthesis filtering. there is no memcpy or memmove used in the relevant parts. so it is of interest how different memory concepts scale
20:33:58 Quit flydutch (Quit: /* empty */)
20:34:04kugelmy intention for test_mem was to check if something is wrong with the fuzev2 actually :)
20:37:32pixelmaBuschel: so you were working on test_mem? Could you look into the issue why "boosted"/"unboosted" often states the wrong thing? Looking at the displayed clock and related speeds seems to be correct but the words don't match
20:37:45kugelthat was one mutex unlock too much, now buttons work again
20:38:09pixelmaon my M5 that is at least
20:40:13Buschelpixelma: that's the same on my machine (iPod 5.5G). seems like there is a delay between triggering the boost ("boosted" is displayed) and the clock change itself. not too interesting though: just wait until the boosted clock is displayed and everything is fine :o)
20:40:50Buschelpixelma: good enough for a tes plugin :o)
20:41:16pixelmaI mean a mismatch between the words "boosted" and "unboosted" and the clock speed displayed along with it
20:42:55pixelmaand yesterday the plugin just stopped on its own without me doing anything and after different amounts of time. I didn't see a pattern
20:45:32Buschelpixelma: I can see the same behaviour regarding boosted and the clock (too much cpu consumed to switck clock? maybe you could just enlarge the get_action timeout?). But I did not experience any plugin exits...
20:47:22 Quit antil33t (Read error: Connection reset by peer)
20:47:28 Join antil33t [0] (
20:49:09 Quit xavieran (Ping timeout: 240 seconds)
20:49:49wodzBuschel: but does codecs use constructs such as for... loop in test_mem or arm asm version or it is just for fun?
20:51:07kugeljhMikeS: ping
20:53:01*kugel is surprised that so many people are interested in test_mem, given that it hasn't been there for so many years :)
20:54:00Buschelwodz: e.g. filterbanks use lots of ldmia to read time domain samples and coefficients, combine them somehow and save a single register at the end. stm is used less often I would say.
20:55:07 Quit stoffel (Remote host closed the connection)
20:55:36Buschelwodz: but maybe we are drifting away from the plugins purpose. If you think memcpy and memmove shall be tested as well -> add them and display the results in new lines.
20:57:30kugeltesting the mem functions is definitely not the purpose i added it for, but it could be useful to see how they perform compared to the real memory bandwidth
20:57:52Buschelkugel: _that_ is indeed a valid point
20:58:42 Quit MethoS- (Read error: Connection reset by peer)
21:00:26Buschelkugel: I thought about calculating the wait states for read/write accesses. Might be helpful when estimating cycle times for asm optimizations on different platforms...
21:01:25wodzdoing simple grep for memcpy shows it is used in codecs as well as in other places
21:01:46Buschelwodz: yes, but not in the speed dominating parts.
21:02:18Buschelwodz: let's just stop this discussion. not worth the time.
21:02:34 Join xavieran [0] (
21:02:51wodzsure. I will prepare the patch to add mem* functions to test_mem in spare time
21:04:41 Quit komputes (Ping timeout: 268 seconds)
21:04:51 Quit n1s (Ping timeout: 252 seconds)
21:05:08Buschelok, keep up the good work!
21:05:51 Join Stephen__ [0] (~S@
21:14:06 Quit JohannesSM64 (Quit: WeeChat 0.3.3-dev)
21:14:08 Join n1s [0] (~n1s@rockbox/developer/n1s)
21:15:41 Part watto
21:19:18 Nick evilnick is now known as sexyduckface (~0c140464@rockbox/staff/evilnick)
21:19:27 Nick sexyduckface is now known as evilnick (~0c140464@rockbox/staff/evilnick)
21:26:24 Quit anewuser (Quit: for SELL 2 by the price of 1 now!)
21:33:40 Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
21:48:50 Quit Jaykay (Ping timeout: 252 seconds)
21:48:51***Saving seen data "./dancer.seen"
21:59:22 Quit moos (Ping timeout: 252 seconds)
22:00:19 Quit kugel (Remote host closed the connection)
22:00:47 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
22:09:15 Join Xerion [0] (
22:12:33 Join m0ar [0] (
22:12:39m0arHello guys
22:13:09 Join M3DLG [0] (
22:13:13m0arHow's the development for the Clip+ proceeding? I'm very new, so pardon if it's a stupid question :)
22:14:11m0arOh, thanks
22:15:21m0arDoes anyone in here run a boxed Clip+? :3
22:16:06Stephen__yes there's a few of us who do.
22:16:39Tornea boxed one?
22:16:41m0arIs it hard to get running? I've never tinkered with a portable audio device
22:16:48Tornetraditionally you take it out ofthe box before running it
22:16:51m0arTorne: Yeah, a rockboxed clap+
22:16:56m0arOh you ;D
22:17:13*evilnick claps
22:17:16 Join smumey [0] (~smumey@
22:18:04Torneok that was a joke but seriously i didn't realise what you meant by boxed
22:19:01 Quit bmbl (Quit: Bye!)
22:19:39m0arSome personal thoughts, what's the biggest reason for you running a rockboxed device?
22:20:18 Quit hebz0rl (Quit: Ex-Chat)
22:20:20wodzit is question for the community channel rather
22:20:46m0arOh, I'm sorry then
22:20:56m0arStephen__: Thanks sherloock, but I asked for personal opinions
22:21:24Tornebecause otherwise it's hard to test the code i'm changing :)
22:22:04Stephen__The wiki is wrote by users so that's the reasons people would use it. any of the reasons people will state are there.
22:22:23Stephen__unless you just want because it's better than the OF
22:22:39m0arOh, thanks
22:22:50Bagderthe main reason is all the money and fame it brings with it!
22:22:57m0arBagder: +1
22:23:13m0arIn this table, USB is NO. Meaning?
22:23:20Tornemeaning usb doesn't work?
22:23:23evilnickm0ar: Being able to drag and drop (if you've been used to using iPods) is a very attractive reason
22:23:31m0arevilnick: nom
22:23:40evilnickAs is the focus on things being based on standards
22:23:57gevaertsm0ar: no usb means you have to use the original firmware for that
22:24:00m0arevilnick: Even tho I've never used any player that didn't support it, that'd be a stupid buy
22:24:11m0argevaerts: For transferring music? :<
22:24:42Bagderm0ar: it really isn't a very big deal
22:24:56m0arSeems like :/
22:25:01 Quit wodz (Quit: Leaving)
22:25:13BagderI don't think you understand it then
22:25:24m0arFeel free to enlighten me
22:25:54BagderI did: it's not such a big deal ;-)
22:26:03gevaertsWhat isn't? ;)
22:26:03m0arI think being able to transfer music to your device is what's called a "key feature"
22:26:03evilnickYou switch to the OF to transfer music, then back to RB to listen to it.
22:26:07 Quit Rob2222 (Read error: Connection reset by peer)
22:26:07Bagderm0ar: you can
22:26:12m0arAh, it's that easy to swap?
22:26:18Bagdervery easy
22:26:40 Join Rob2222 [0] (
22:26:46Bagder(and the text on that wiki page explains it a bit too)
22:26:49m0arJust read that I've got a probability of completely breaking my clip+, sound like fun to change firmware often :D
22:27:07Bagderthere's only really a one-time risk
22:27:17m0arDo explain, please
22:27:29Bagderthe risk is when/how the bootloader is installed
22:27:30evilnickm0ar: Once you've installed rockbox, it's really just like picking which desktop OS to launch
22:27:47m0arI can seriously dualboot my player
22:27:51evilnickExcept without a menu and having to hold buttons whilst booting
22:27:57Bagderm0ar: and for most uses, you will never even install the bootloader
22:28:00m0arThat provides such a geekstamp
22:28:07Bagderuh, darn, never again I meant
22:28:27*Bagder signs up for typing class
22:28:29m0arSo what may cause the installation of the bootloader to fail?
22:28:34Tornevery little
22:28:39 Quit Unhelpful (Quit: No Ping reply in 180 seconds.)
22:28:40*m0ar slapts Badgers fingers
22:28:42Tornei don't think anyone has bricked a clip+ yet, have they?
22:28:46 Join Unhelpful [0] (~quassel@rockbox/developer/Unhelpful)
22:28:52Bagderyes that Badger guy should have it
22:28:59m0arGood, but what may cause a breakdown?
22:29:10Tornecutting power while it's installing
22:29:16Torne..that's probably it
22:29:25Torneand it takes a couple seconds, so it's not exactly a large window
22:29:34Tornewhat did you expect?
22:29:38m0arWell, I'll have it charged
22:29:49Torneso the risk is very small
22:30:13Tornethe bootloader installer only works with original firmware updates which we have tested and included the checksum of
22:30:22Torneso there shouldn't be *any* risk of it patching ht ecode wrong
22:30:30m0arJudgind from the wiki from the viewpoint of a total n0oBBb: The smallest inferference will render your Clip+ useless.
22:30:53m0arYeah, that's why I came in here :D
22:31:06Torneoh, you might render it useless at any moment, but not *permantly* :)
22:31:17m0arI meant permanently.
22:31:29m0arAs in:
22:31:35evilnickm0ar: Which section are you referring to?
22:31:38Tornevery few targets have any measurable risk of bricking unless you do something really stupid
22:31:41evilnick(on the wiki)
22:31:50Tornesome of theunstable targets might randomly reformat themselves or corrupt the data or whatever
22:31:54m0arthere's always a risk of BRICKING your device PERMANENTLY., which repeats again and again
22:31:57Tornebut it will just kill all your music and force you to reinstall rockbox
22:32:15Tornegenerally there's no risk of bricking the device once you have installed the bootloader
22:32:18Torneunless you install a new bootloader.
22:32:19m0arWell, good :)
22:32:22evilnick"Even though the bootloader installation and rockbox itself are tested, there's always a risk of BRICKING your device PERMANENTLY."?
22:33:06m0arevilnick: Well, the red text warning for a takedown lurks in a few places more. The fact that it's well-repeated made me wonder :)
22:33:19Bagderwe want the warnings to be big and bright so nobody comes down and say we didn't warn them, for the odd case where something actually does go wrong
22:33:49m0arProbably some guy that gets tired of waiting for the "firmware update" and just forces it off
22:34:05m0arA question, why isnm't it possible to reinstall a standard bootloader if installation fails?
22:34:17Tornebecause the firmware is what installs the firmware, on a lot of players :)
22:34:36Torneif you break it there's nothing left to install another one with
22:34:39Bagderif installation fails, the device is dead
22:35:12Bagderwell, at least it some cases ;-)
22:35:23m0arWhen I get my player I'll come in here and nag you guys while I do this, minimizing the chances of me failing miserably
22:35:51Tornewe'd rather you didn't
22:35:53evilnickIt *should* be perfectly straightforward to follow the instructions and do it on your own
22:36:16Tornefollow the instructions, and ask us if something is unclear
22:36:26m0arTorne: Yeah, that's what I meant dude
22:37:12AlexPAlso, make sure to read the manual before asking
22:37:18m0arOf course
22:37:20Bagderand we can sing a song to calm him down!
22:37:29AlexPGood idea to do that now - it tells you about dual-boot for instance
22:37:38m0arAlexP: Yeah, I'm on it already :)
22:37:50evilnickDoes the manual exist for the clip+?
22:38:37AlexPevilnick: That is a reasonable point
22:38:56AlexPThat is as it is classed as unusable
22:39:23BagderUnstable actually on that wiki page
22:39:46AlexPIt is unusable on the front page
22:39:55AlexPSo which is it?
22:39:56 Join Strife89 [0] (
22:39:57Tornei expect it didn't get put back
22:40:02evilnickAnd some bits of the manual do exist:
22:40:19evilnickPresumably those parts that are the same as the clip
22:40:30AlexPTorne: Ah, it got prematurely made unstable?
22:40:43Tornethe change to the front page got reverted
22:40:44AlexPI'll change the wiki page back
22:40:47Tornemaybe the wiki wasn't revised
22:40:57Tornethough now that writes work maybe it *should* be unstable again
22:41:16Tornei haven't been paying enough attention to comment though
22:41:18m0arIs background noise almost completely dependant on firmware?
22:41:41Tornemost players i would suspect hardware
22:41:52Tornebut there are instances where we have less noise, i'm sure
22:42:06m0arOkay, just wondered since that's a feature listed in the Firmware Feature Comparision Table
22:42:08AlexPTorne: I'm putting the wiki to unusable - if they deserve to be promoted then they can be at the same time
22:42:29Tornem0ar: well, there are some players where we know the OF has noise problems and we don't :@)
22:42:33Tornethat's why it's there
22:42:47Tornebut on most mp3 players that exist any noise problem there is, is probably down to hardware
22:42:55Tornebad design of analogue paths
22:43:37m0arAh, rite
22:49:54 Quit domonoky (Read error: Connection reset by peer)
22:53:04Tornehaha, having just had this conversation the latest post in the forum is from someone who appears to have bricked their clip+ during bootloader install
22:53:31m0arLink :D
22:55:21 Quit Buschel (Ping timeout: 260 seconds)
22:58:45 Join bluebro [0] (~4d2c4024@gateway/web/freenode/x-pwzyzmxjamwwnoug)
22:58:59 Quit bluebro (Changing host)
22:58:59 Join bluebro [0] (~4d2c4024@rockbox/developer/bluebrother)
23:00:00FlynDiceTorne: can you point me to the message
23:00:40Tornegevaerts already responded with "have you tried resetting it" which is probably sufficient :)
23:00:43Tornei just found the timing amusing
23:03:38Stephen__is there a link to the current source ? as opposed to using svn ?
23:04:12bluebroStephen__: no, we don't provice an svn tarball anymore. There isn't much point in doing so
23:04:52Stephen__ > should be changed where it says: Daily builds, voices, fonts and source
23:04:59Stephen__I just grabbed 3.5.1 anyways
23:05:33bluebrogood point.
23:05:40*bluebro looks around for Bagder
23:06:33*Bagder jumps out from the shadows and says "g'day sir, you called?"
23:08:11 Nick Ypsy is now known as YPSY (
23:09:58 Quit Xerion (Quit: )
23:10:29bluebroBagder: any thoughts on removing the "sources" line from the daily build page? (The "Daily builds, voices, fonts and source" on that is)
23:11:12Bagdermaybe we should have daily sources...
23:12:46Tornewastes our bandwidth, no?
23:12:58Tornecan we just get ar ead-only http svn mirror instead?
23:13:06Torneso that people who can't access svn because of firewalls can use it?
23:13:10 Quit evilnick (Quit: Page closed)
23:13:32Tornewell i say mirror, i just mean a mod_dav_svn on the machine that hosts the svn-protocol svn at the moment
23:14:16bluebroweren't the daily tarballs removed because they caused too much work (those FILES files IIRC?)
23:14:38bluebrobut I wouldn't object to daily sources. I won't be the one to implement that anyway :)
23:15:03Tornesvn not being accessible behind some firewalls is about the only sensible reason why telling people to use svn isn't sufficient, imo
23:15:12BagderI agree
23:15:16Torneif people are going to compile the source, then they alrady need a whole bunch of stuff
23:15:22Tornerequiring that they also have svn is not exactly arduous
23:15:35Torneand http svn is not hard to set up as long as you only support reads
23:15:42Tornesince you don't need any access control then :)
23:15:44joeygwe could move to git and gitweb ;)
23:16:17bluebroand what's the gain with git in that area? ;)
23:16:34Bagderit's not even shorter than svn! ;-)
23:16:50bluebrothen we'd need to move to hg!
23:17:10Bagderoh indeed, I'm now convinced!
23:17:27bluebro(though I liked git much better than hg when trying)
23:19:01joeygyou can use git over http i believe
23:19:07joeygso no firewalls to worry about
23:19:18Torneyou can use svn over http
23:19:22Torneso that's not really an argument
23:19:59joeygyou suggested svn was not accessible behind some firewalls - in what way?
23:20:18Torneat the moment we expose the repository only through the "native" svn protocol, on the port number usually used for that protocol
23:20:30Tornesome people's networks only allow connections to fixed ports
23:20:37Torneproxy-only internet access and all that
23:20:48joeygi gotcha
23:20:48bluebrogit is completely different than svn. While I like it pretty much I don't see much gain for the project when switching to git. It we were still using cvs the things would be different ...
23:21:04Tornemod_dav_svn exposes svn repos through WebDAV, which then works on port 80 same as everything else
23:21:18Tornebluebro: i just ignore everyone and use bzr
23:21:20Bagderbluebro: being a git convert myself now, I *do* see a benefit from using git ...
23:21:29Torneand if the project switched to git i would continue to use bzr :)
23:21:45joeyg:) i use git-svn with no trouble at all
23:21:58joeygthough im exploring the s9 so im off in my own little world..
23:23:08bluebroBagder: of course there are benefits, but with git-svn everyone can chose :)
23:23:29*bluebro didn't like the branching handling of bzr
23:24:24Tornebluebro: it's just a UI thing
23:24:41Tornegit has a totally different way of working to all the others (which are all much the same as each otehr)
23:24:52Tornebut they all do almost *exactly* the same things in almost exactly the same way really
23:25:27bluebrowell, I wouldn't call this "every branch needs its own folder" a UI thing compared to git's "only one work area"
23:25:35Tornebut they don't.
23:25:40TorneAnd it *is* a UI thing
23:25:43bluebrobut yeah, from the functional point of view it's pretty much the same
23:25:49Tornethe underlying data is identical
23:26:10Tornealso, my rockbox folder has 20+ branches and only four folders :)
23:26:52Tornealso also, you can work exactly the way git works if you want
23:26:57linuxstbjoeyg: I've just read the recent posts in this thread - - you can just post the URL I gave to the windows binary. Which BTW, I cross-compile on Linux using the mingw32 package - just install that and type "make tcctool.exe"
23:26:59Torne(mine is using a nonstandard plugin, but hey)
23:27:11Torneyou just make the branches without workdirs, and then do a lightweight checkout to someplace else
23:27:18Tornethen you can switch the branches of that checkout, exactly as with git :)
23:28:04bluebrowell, I really found that confusing. Especially the explanations the UI gave me (this tortoise-like interface that comes with bzr in my case)
23:28:37joeyglinuxstb: thanks, i'll post that link when i get home from work
23:28:48joeygand thanks for the mingw32 tip, i had no clue that's what that was
23:28:57Tornei'm not sure why branches being directories is confusing, though, still.
23:29:04Tornei find git's way *extremely* confusing
23:29:10Tornei much prefer being able to see it there in my filesystem
23:31:28joeygive always liked git's way
23:31:42joeygjust typing 'git init' on every new project, then 'git branch' to see branches
23:32:28 Quit bluebro (Ping timeout: 252 seconds)
23:34:46 Quit CGL (Quit: Soy spammero ---- \m/ d(>.<)b \m/)
23:35:37Tornejoeyg: but i build anything up to half a dozen branches of rockbox at a time
23:35:49Tornejoeyg: having them in seperate directories to start with makes that trivial
23:36:09CIA-5New commit by alle (r25958): Add the cyrillic 'yo' to 12-Adobe-Helvetica font
23:37:02joeygthough, you could add 'git branch' instructions to your build script and get a similar effect
23:37:26joeygi agree that's a silly pain though
23:38:27Tornei understand how to work with it, i just don't see where it's an *advantage*, excpe tit takes slightly less disk space
23:38:31Tornedisk space is cheap ;)
23:39:46Torne(and as i noted above you can not hav ea working tree checked out in each branch, if you want, and just keep any number of seperate trees)
23:40:23joeygit lets git store branches as diffs of each other and with its linked-list type structure
23:40:30joeygwhich is a fair bit faster and simplifies merging
23:40:35Torneer, bzr does that though
23:40:41Tornei only have one copy of all my revisions
23:41:09Tornethe seperate branch folders are just a tiny metadata file that says which commit is the tip of that branch
23:41:13Torneand then optionally a workdir
23:41:28Tornethey're in folders because the usual case is to have a workdir, so they need to be in one anyway
23:41:52*gevaerts finds this all very interesting, but...
23:41:59Torneokay yeah, this is kinda offtopic ;)
23:42:07joeygalright :)
23:42:14TorneBagder: are you the one to pester about getting http access to svn, though?
23:42:20Tornereturning to the actual point :)
23:42:33*Bagder whistles as he's slowly hiding behind gevaerts
23:42:41AlexPThat would be useful :)
23:43:10Bagderme or zagor would be the guys to pester, yes
23:43:39Bagderbut since zagor is younger, you should poke on him!
23:43:47AlexPhaha :)
23:43:50scorche|shand he isnt shunning devcon!
23:44:06Bagderright, you'll reach him easily then!
23:44:21Tornewell, i don't personally need it.. it's just that *sometimes* people have the valid reason that they want to ahve a source archive because they can't get to svn
23:44:27Torneso we should maybe fix that
23:44:34Torneso we can just accuse them of being lazy instaed ;)
23:48:53***Saving seen data "./dancer.seen"
23:53:05 Join bluebro [0] (~4d2c4024@gateway/web/freenode/x-erwydudfoetocpvd)
23:53:16 Quit bluebro (Changing host)
23:53:16 Join bluebro [0] (~4d2c4024@rockbox/developer/bluebrother)
23:53:25 Quit bertrik (Quit: Leaving)
23:54:28 Join anewuser [0] (anewuser@unaffiliated/anewuser)

Previous day | Next day