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 2007-10-21

00:00:15kugeli wonder why. does the flash memory die earlier with dircache?
00:00:22mrkikoIf someone can help me, I'm available - I will not repeat this message so no one will annoy. Thank forreplies.
00:00:34Slasherirasher: how about displaying a message that reverse scan was not performed when entering into tagcache screen?
00:00:43ddaltonmrkiko: I sometimes find the same problem.
00:00:57ddaltonactually 5 chrashes last night in half an hour...
00:01:12mrkikoddalton: oh... what player do you use?
00:01:25Slasherikugel: nope, it's just so fast that there isn't much benefit from dircache
00:01:27mrkikoddalton: wowowowo
00:01:45Slasherino spin-up times, fast seek etc.
00:01:47mrkikothis is my happy night! :)
00:01:47rasherSlasheri: I agree with dircache not being terribly useful on non-flash, but.. a) It's slightly faster than reading the FAT b) some people are using multivolume with hdd.
00:01:48 Quit Mouser_X (Nick collision from services.)
00:02:03ddaltonrasher: How do I make the .talk clips for the db using your perl script?
00:02:06mrkikoddalton: how did you reset it? I need help from someone.
00:02:28ddaltonmrkiko: by pressing the button on the side of the unit...
00:02:38 Join Mouser_X [0] (n=someone@
00:02:40rasherddalton: well you run the script and pass it thw options it needs, basically. What's not working?
00:02:45mrkikowith something like a pencil?
00:03:00kugelbut there isn't a real drawback of dircache on flash, or?
00:03:00rasherddalton: also note that Database doesn't support it yet, so the clips are rather useless on their own
00:03:10Slasherikugel: shouldn't be
00:03:17ddaltonrasher: I can't work out what options I give it. Can you give me an example
00:03:21Slasherikugel: so far that patch looks ok
00:03:24rasherkugel: it uses up RAM, but other than that no, just not very many advantages
00:03:33ddaltonI want to see how the .talk clips are named so I can try and make them work in the db...
00:04:03ddaltonmrkiko: I used a pin because it has to be very small
00:04:06rasherddalton: ./ -D -t="flite" -T="" -e="lame" -E="" -l="english" /path/to/player/root
00:04:10kugelrasher talked about a hot swap issue with dircache, how about that?
00:04:29rasherkugel: that's what this patch addresses...
00:04:30ddaltonthanks I will try it
00:04:38Slasherikugel: that patch seems to rescan disk when hotswap event is issued
00:04:56Slasheriso it should work too
00:05:15kugeli see
00:05:33mrkikoddalton: have you submitted a bug report? I looked but I can't find one.
00:06:02ddaltonno im not sure how to fix it.
00:06:20pixelmaI thought the real issue wasn't dircache on flash targets bur with the multivolume - and different amount of the ram it needs
00:06:48ddaltonis there a sighted user here than can try something for me?
00:07:09ddaltonmrkiko: Let me see if I can reproduce it.
00:07:39mrkikoddalton: if you want you can reproduce it simly interrupting the voice ...
00:08:39Slasherii think it's ok to commit that dircache patch unless there is anything against it
00:09:17rasherSlasheri: I think you're pretty much the only one who could object against it. Which is a bit of a shame
00:09:40kugelSlasheri: you might contact the patch maker
00:09:45Slasheriwell, i will do something for that tomorrow then
00:10:05kugelhe could be a help
00:10:48ddaltonmrkiko: oh ok just a press of up will do?
00:11:45mrkikoddalton. and... how do you tell when the battery is discharged at all? My reader turned off
00:11:57mrkikobut I would be noticed first (how it happens in the original firmware)
00:12:08mrkikowithout checking the place in themenu
00:12:29kugelSlasheri: the patch is out of sync atm (but only 1 hunk fails), i could upload a sync'd version if you want
00:12:52ddaltonmrkiko: what about the battery? I use a patch that warns me.
00:13:15Slasherikugel: it applied fine without errors for me
00:13:24mrkikoddalton: are you the patch author?
00:13:24ddaltonmrkiko: where do you see this problem? What menu. What key press? A short or long press?
00:13:32Slasherito the latest svn
00:13:36kugelon my sansa there was a hunk failed
00:13:46mrkikoddalton. ...
00:13:55ddaltonmrkiko: which patch? FS #7910?
00:14:07mrkikoddalton: a sequence of ten short pressed in a spoken menus does for me...
00:14:23ddaltononce second...
00:14:29*ddalton Goes and tests
00:14:30kugeli meant, i am building for sansa, which isnt necessary
00:15:16kugelbut nevertheless i had a hunk failed (with the last posted version though)
00:15:37mrkikoddalton. I meant the patch that warns you about the battery being exhausted. I asked you who was the author. And why not just commit it upstream? I'm not a rockbox expert so this is just a question...
00:15:59 Quit w1ll14m ("CGI:IRC (Ping timeout)")
00:16:22ddaltonmrkiko: the one that says the battery level at 50, 30 and 15% is by me yes.
00:16:30ddaltonI am hoping it will be committed.
00:16:40Slasherirasher: i will fix the tc debug screen so that it always shows the current file in realtime
00:16:44ddaltonbut should probably remove the charging stuff.
00:17:19ddaltonmrkiko: I reproduced. Can you give me some time to work on a fix?
00:17:27ddaltonIf I can...
00:17:33mrkikoddalton: congratulations! May be some day I'll be able to program in projects such as rockbox...
00:17:48ddaltonmrkiko: can you program in c?
00:17:58mrkikoddalton: yesss
00:18:23mrkikoddalton. but, for example, i don't know how to debug this problem, since when my reader is crashed I can only notice that it's crashed...
00:18:51mrkikoddalton: d you understand what I mean? I would be very happy if an ncurses-based emulator exist, but for now the SDL one is the only one I see
00:19:19*mrkiko avoided by miracle to use capital letters, avoiding to be kicked :)
00:19:37 Quit ompaul (Client Quit)
00:21:28rasherSlasheri: I don't know if that's strictly necessary - just staticly showing the last one at the time the screen is entered would probably work as well, but it's of course slightly nicer
00:21:38ddaltonmrkiko: I am blind and I am just learning c. so you could probably code on rb. Just got learn where everything is and
00:21:42ddaltonwhat functions to use.
00:21:55ddaltonThe hard thing is testing but with vice that's ok.
00:21:57rasherSlasheri: It'll also help expose problems in tagcache and/or the metadata readers
00:22:07ddaltonThen sighted devs usually tell you if something is wrong.
00:22:17ddaltonBut first write some basic patches
00:22:30ddaltonalso there is a simulator
00:22:33Slasherirasher: indeed, i will just make it synchronous
00:23:24ddaltonmrkiko: Hmmm I can't seem to reproduce it very easily.
00:23:38mrkikoddalton. I developed several programs in C so i may know it (or at least it's basic concepts). My problem has always been to find out what functions of the 'kernel' or the 'core firmware' I should use.
00:23:55ddaltonIf I write a couple of patches for you can you see if the problem still occurs?
00:24:01mrkikoddalton: can you go on the fm recoder -> virtual keyboard?
00:24:27ddaltonmrkiko: where you add a preset and it gives you all the letters?
00:24:29mrkikoddalton. yes! I would be very happy to help you in any manner I can.
00:24:45ddaltonsure hang on one minute
00:25:11mrkikoddalton: is this a question?
00:25:49mrkikoddalton: I'm a little obsessed aniway by crashes: I "love" my iRiver and resetting it make me displease a lot! :)
00:26:25mrkikoddalton. are you the voice-layer writer?
00:26:28ddaltonmrkiko: Ok im there what do you want me to try?
00:26:52ddaltonwhat do you mean by "voice-layer"?
00:26:54mrkikoddalton. then yo may try settings -> record settings
00:27:17mrkikoddalton. voice-layer = the set of instructions regulating voice reproduction
00:27:52mrkikoddalton: The window where you can choose things such as bitrate and so on...
00:28:02mrkikoIf you scroll the menu two or three times it will crash...
00:28:06rasheramiconn: share your displeasure - it helps
00:28:47ddaltonmrkiko: hmmm I can seem to make it re occur. ill will write you up a patch. It will go really really slow but then you can just see if it still occurs.
00:28:50ddaltonAre you blind
00:28:59*ddalton for got the question mark
00:29:30amiconnI need to find the correct init sequence for the small H10's lcd, but the OF puts everything into structures (unlike the apple OF's), hence I can't seem to dig that down :(
00:30:23*ddalton reproduce in the display menu...
00:30:33ddaltonI think it is because of the long lang entry.
00:30:33mrkikoddalton. yes
00:30:34 Part keanu|away
00:30:57 Join keanu|away [0] (
00:31:05mrkikoddalton. how will you figure out where the problem is?
00:31:07ddaltonmrkiko: re the voice layer question. I didn't write the voice ui but wrote some patches for it.
00:31:15ddaltonsdoyon has done a lot more than me though
00:31:56ddaltonmrkiko: Well I am going to try with action_none. If that works then I need to figure out how to wait on a button release
00:32:07ddaltonactually that doesn't make sense.
00:32:14ddaltonthey are all button releases...
00:32:22ddaltonUnless you hold of course
00:32:33ddaltonmaybe detect how long it was since the last press.
00:32:57ddaltonmrkiko: have you got your dev environment setup?
00:33:12mrkikoddalton. do you use a remote controller for your H320?
00:33:31ddaltonmrkiko: sometimes
00:33:34mrkikoddalton: no, but I may set up one fastly
00:33:42ddaltonthe one that came with on the leed...
00:33:55 Join mo-seph [0] (
00:34:09ddaltonmrkiko: I am not sure how long it will take but does someone a link for getting setup?
00:34:17ddaltonyour on linux?
00:35:06mrkikono ... don't worry
00:35:09mrkikoI know how to do this
00:35:20mrkikoat least in theory
00:35:53mrkikoI'm checking out svn
00:35:54ddaltonmrkiko: ok setup your compilers and stuff then
00:36:24mrkikoShould i use the devscript aniway or does a debian package exist with those compilers?
00:36:28 Quit keanu|away ("Connection eaten by dog")
00:36:28 Quit courtc (Read error: 104 (Connection reset by peer))
00:37:13 Quit hannesd ("Client suicide")
00:37:38scorchemrkiko: is the easiest way to do it
00:38:05ddaltonso is that all I need to do when I get fedora? just run that scrpt?
00:39:25 Join courtc [0] (
00:39:56 Join keanu|away [0] (n=chatzill@unaffiliated/keanu)
00:40:57mrkikoInfact I'm running that script
00:41:20mrkikothe problem is that I'm using UMTS - not so fat
00:41:26ddaltonmrkiko: this isn't the final way but if we find out it works we know we need to not do anything if there is more than one key press in .5 seconds.
00:41:38ddaltonwait I might have a simpler idea
00:42:51mrkikoddalton: ...
00:43:39mrkikoddalton. but... if yopu interrupt the voice and press another key, shouldn't the system just start saying the other thing and have no effect?
00:43:46mrkikoI don't know if I'm explaining myself
00:43:58mrkikoHere the pc is compiling
00:44:07 Join Gnu47 [0] (
00:45:29ddaltonmrkiko: I better go but I will be back in about an hour. My problem is that I can't reset a bool variable if we have pressed again.
00:46:22ddaltonBut I might just put the code in action_none and see what happens. Still need a counter. And will be a long delay. But if it works then we can probably figure it out.
00:46:35ddaltonnot a counter a bool variable...
00:47:13mrkikoah... I understood
00:47:15ddaltonmrkiko: talk only if there hasn't been a keypress for 0.5 seconds.
00:47:21mrkikoI'm compiling gcc again
00:47:49mrkikoddalton. you're simply a geinius :)
00:48:02 Quit davina (Remote closed the connection)
00:49:22rasherddalton: that seems like a cludge around whatever the real problem is
00:49:51ddaltonmrkiko: im not sure exactly maybe do_shutup and talk_disable then after 0.5 seckons then just restore everything and talk.
00:50:04ddaltonif there hasn't been a key press.
00:50:22 Quit freqmod (Read error: 104 (Connection reset by peer))
00:50:30ddaltonI need to fine though. 1. How to do detect a key press 2. Where the code is for scroling through menus.
00:50:50*ddalton Will be back
00:51:31ddaltonBTW before I go can someone see if they can reproduce this problem with voice disabled?
00:52:15 Quit mo-seph (Read error: 110 (Connection timed out))
00:52:30 Join webguest23 [0] (i=4cd40747@gateway/web/cgi-irc/
00:53:49 Join webguest19 [0] (i=4cd40747@gateway/web/cgi-irc/
00:53:49 Quit webguest23 (Client Quit)
00:56:09 Quit webguest19 (Client Quit)
00:56:39 Quit courtc (Read error: 104 (Connection reset by peer))
00:58:02 Quit Thundercloud (Remote closed the connection)
00:58:14mrkikoddalton: probably this will not happen...
00:58:38 Join courtc [0] (
01:01:28mrkikoWhile are some files regarding z80 being compiled for my iRiver ? :)
01:02:29linuxstbEither Rockboy, pacbox or zxbox - emulators.
01:02:51 Join midgey [0] (
01:03:28 Join feindbild [0] (
01:03:45feindbildhi =)
01:04:35mrkikofeindbild: hi
01:04:42mrkikolinuxstb: ah...
01:04:44 Nick idnar_ is now known as idnar (i=mithrand@unaffiliated/idnar)
01:05:20feindbildwhere can I browse the microsd on the sansa e200? (latest rockbox build)
01:05:51n1sfeindbild: it should be visible in the file browser root as <microSD> iirc
01:06:08feindbildn1s: no =(
01:06:44mrkikoapart blind person: how many of you are interested in including real-tiem speech in rockbox? Would be in your opinion too heavy for an embedded CPU?
01:07:58n1smrkiko: I think it would be great, an can probably be made fast enough, there is an espeak plugin in the tracker but there are license problems (gpl v3 vs v2) :-/
01:08:01mrkikoddalton: when you're back (obviously) can you tell me if you found a trick to turn the player off using the remote controller? It would be very handy but I can't find this info on the manual.
01:08:15linuxstbmrkiko: There's lot of interest in it. There was a Google Summer of Code project, but the student didn't do any work... Plus the plugin n1s just mentioned.
01:09:26mrkikoI seen the plugin in the tracker but, noting no-one is working on it- I was thinking the project to be 'died'
01:09:39mrkikorockbox is gplv3?
01:09:43mrkikoand espeak v2?
01:09:51linuxstbNo, Rockbox is gplv2
01:10:43mrkikolinuxstb: if I'm not wrong, GPL allows generally you to pick a version of espeak made under gplV2 and use it under this license...
01:11:31linuxstbespeak is gplv3, so we can't use it - unless we went back to an older version before it changed to v3.
01:11:33rashermrkiko: yeah, the last gplv2 version of espeak can certainly be used
01:12:32 Join noztrilz [0] (
01:12:40mrkikoI tried flite - not so bad
01:12:45mrkikoIt's a lite version of festival...
01:13:30feindbildn1s: could it be that the microsd isn't allowed to be partitioned?
01:13:50***Saving seen data "./dancer.seen"
01:14:07rashermrkiko: I still think flite is a lot slower than espeak (but I'm basing this on nothing, to be honest)
01:14:09n1sfeindbild: I'm not very familiar with the details but yeah I can imagine that being a problem
01:14:39noztrilzis the sansa release atmo, a finished/complete release?
01:14:59feindbildnoztrilz: USB isn't working yet
01:15:19Zagornoztrilz: software is never finished
01:15:23noztrilzis that all?
01:15:38mrkikorasher: ... you're right
01:15:54noztrilzwhat part of the USB isnt working?
01:16:19feindbildnoztrilz: as far as I know: non. neither charging nor file transfer ..
01:16:43rasherCharging is sort of working, but slow
01:16:45noztrilzi just got the latest release and data transfer works fine
01:16:58noztrilzis that because its using old sansa firmware
01:17:12mrkikohow did rockbox interfere with battery charging?
01:17:17n1smrkiko: also espeak seems to still be developed, its 2 years since the last flite release... (not that frequency of releases is any measurement of activity, but it seems quite dead)
01:18:28mrkikon1s: aniway the activity of rockbox impressioned me...
01:18:34noztrilzdoes anyone know where i can get any more (apart from rockbox website) sansa themes/wps?
01:18:46n1sand we haven't released in years :-)
01:20:13rasherEek, it's been 2 years. At this rate we're going to have a worse track record than Debian
01:20:29 Join AndrewJ [0] (n=KeyLime@
01:20:48n1srasher: they are at least working toward a release ;)
01:21:37AndrewJanyone have any idea when embedded album art might work?
01:22:04noztrilztombstone was fuckin awesom
01:22:17ddaltondoes anyone know how to detect a key press in 0.5 seconds?
01:22:30noztrilzmy apologies
01:23:27ddaltonmrkiko: I guess I will have to wait until JdGordon comes on because he knows about buttons
01:23:35Zagorddalton: button_repeat
01:23:41 Quit midkay ("Leaving")
01:23:41noztrilzso, anyone know any good themes or wps released for sansa?
01:23:44ddaltonand I have know idea how to detect key press in 0.5 s
01:24:25ddaltonZagor: how can I tell in menu.c if it has been pressed released then tapped again in about under a second
01:24:29mrkikoddalton. don't worry
01:25:02mrkikoddalton. the important thing is to note what's happening...
01:25:04ddaltonmrkiko: did you get your environment setup?
01:25:26Zagorddalton: why do you want that in menu?
01:25:44mrkikoI built rockbox already once now to try it :)
01:26:27ddaltonZagor: well if you tap it once it can be delayed for 0.3 and if you tap it twice it won't talk until no activity for 0.3 s. I guess 0.5 would be to long.
01:26:35ddaltonand to stop chrashers
01:26:53ddaltonmrkiko: it worked?
01:27:28rasherddalton: The real solution is to find out why it crashes - not to work around it with a nasty hack
01:27:40ddaltonrasher: ok.
01:27:59mrkikoddalton: yes, it worked
01:28:03ddaltonI think it is voice been interupted. So with voice off it shouldn't chrash
01:28:15ddaltonmrkiko: good.
01:28:35rasherDoesn't voice already work fine with being interrupted before it's done talking? I've never had problems with that in my small tests
01:28:43ddaltonrasher: can you try something?
01:28:45mrkikobut I'm getting a little tired - I'm up since 6 AM ...
01:29:08ddaltonwell it does for long menu items on h300. (A quick key press after another one.)
01:29:18rasherddalton: if you'd specify "something", I can answer you
01:29:20ddaltonbutton_repeat its fine
01:29:42ddaltonrasher: ok well do you know the graphic eq? can you tell me the four options?
01:30:04 Join linuxstb_ [0] (
01:30:30rasherddalton: I'm not sure what four options you're talking about
01:31:10ddaltonrasher: ok in that screen what happens when you arrow around it.
01:32:30 Quit midgey ()
01:32:34ddaltonrasher: so if you press down what options do you have?
01:32:59 Quit AndrewJ ()
01:33:17 Join barrywardell [0] (n=barrywar@
01:33:25rasherddalton: (this is the keys for sansa) There are 5 bands which you can select with up and down. And for each band there are three parameters you can select by pressing right (it wraps): Gain, cutoff frequency, Q
01:33:45rasherIt's basically the exact same thing as the Advanced equalizer settings
01:34:10rasherexcept compressed into a single screen, rather than in menus
01:35:00 Quit linuxstb (Nick collision from services.)
01:35:02 Nick linuxstb_ is now known as linuxstb (
01:35:30BigBambimatt bery
01:35:51rasherI don't think there's much point in making the graphical eq accessible.
01:36:14rasherIt's pretty much guaranteed to be more confusing than the advanced eq settings menu
01:36:24 Join sdoyon [0] (
01:36:55rasheror at best, it'd be pretty much identical
01:36:56 Join JdGordon [0] (
01:37:24JdGordonsdoyon: hey
01:37:44 Quit n1s ()
01:38:08rasherddalton: I can't make voice crash by interrupting it by moving fast in the menu. Something else is going on.
01:38:30mrkikorasher: what player are you using
01:38:54 Quit ender` (" In the beginning the Universe was created. This has made a lot of people very angry and been widely regarded as a bad move.")
01:39:09sdoyonJdGordon: hey!
01:39:27JdGordonwhat happened to your voice callback patch? was i committed yet?
01:40:04sdoyonJdGordon: thought to do that tonight actually. Why?
01:40:20mrkikorasher: no... the problem is not the voice crashing; the entire player go mad :) :)
01:40:27JdGordonbecause the new list api i put in last night was going to use it, but it isnt in svn yet
01:40:50 Quit bertrik ("sleep")
01:41:18rashermrkiko: my point is that it's not as simple as voice can't handle being interrupted, since that's working just fine on my player. Something more sinister is obviously going on.
01:41:28sdoyonJdGordon: saw the commit log but haven't checked the code yet.
01:41:51mrkikorasher: yes...
01:42:03JdGordonok, well if you commit the callback stuff, ill update that new api for it
01:42:57rasherJdGordon: do you know the database menu code?
01:43:05mrkikorasher: it was an exclamation to make you... smile
01:43:18 Part noztrilz
01:43:27JdGordonrasher: tagnavi? or its actual menus?
01:43:50ddaltonsdoyon: does your eq patch voice the graphical eq?
01:43:57Buschelhmm, did anyone of you ever get "internal compiler error: in emit_move_insn, at expr.c:3092"?
01:44:09rasherJdGordon: the code that displays and navigates the tagnavi-generated menu. I'm looking for someone to implement the rockbox-side of talking Database
01:44:18 Quit Zagor ("Client exiting")
01:44:31rasherJdGordon: I've got clip generation done, now I need someone to put code in to play those clips as appropriate
01:44:31JdGordonno, i tinhk i looked at it once
01:44:42JdGordonSlasheri is the man to ask
01:45:25JdGordonwhere are the talk clips? and how are they linked o a string?
01:45:29rasherTried that, sounded like he wasn't sure how it worked either. Maybe he just wasn't terribly motivated
01:45:38sdoyonrasher: I'll be glad to look at it of course.
01:45:39rasherJdGordon: A folder in .rockbox, named after the string
01:46:07rasherJdGordon: sanitized to be legal fat-name and truncated to 150 chars if necessary
01:46:30markunmrkiko: another advantage of espeak over flite is that it has support for many languages
01:46:30rasheror 100, I forget what I set it as
01:46:48markunmrkiko: and it should consume a lot less memory
01:46:55JdGordonsdoyon: is that voice callback patch committable nowish?
01:47:08JdGordonrasher: i dont tihnk what you want is hard to do at all actually..
01:47:14JdGordonjust need to find the right place to put it :)
01:47:16rasherAlternatively, I suppose I could generate a "language clip db", but I think this method works well enough
01:47:25ddaltonrasher: does it create .talk clips for artist, album etc
01:47:29sdoyonIdeally this doesn't need to be too specific to the database stuff. If we can just lookup strings in general. Could be useful for other things, like radio presets perhaps.
01:47:31rasherJdGordon: exactly.. and I've personally got absolutely no idea where to look
01:47:33rasherddalton: yeah
01:47:40mrkikomarkun: right
01:47:50*JdGordon having a look-sie for ya
01:47:52rashersdoyon: sure, and it could replace the current .talk clip method as well
01:47:56ddaltonhmmm that could be useful for sdoyon's id3 patch as well. maybe
01:48:00sdoyonJdGordon: re voice callback, yes, shouldn't take too long.
01:48:02rasherif desired
01:48:10*sdoyon goes to get list voice callback committed.
01:48:28JdGordonthat callback will make things a bit easier actually
01:48:32ddaltonhey sdoyon: did you answer my question? I might of missed your message
01:49:56JdGordonrasher: hows your c?
01:50:22*markun still thinks we don't lose anything by having a LGPL API file to make it possible to have non GPLv2 compatible plugins
01:50:32rasherJdGordon: not so good
01:51:08markunwell, we lose something of course: control over licenses people chose
01:51:10JdGordon:) well the functinos you want to edit are ft_play_dirname and ft_play_filename in tree.c
01:51:33rasherJdGordon: that's helpful.. those are called even in Database mode?
01:51:40JdGordonit appears so
01:51:53rasherThen it should be a walk in the park from there
01:52:14*mrkiko speaking at the phone
01:52:21JdGordonexcept it might need to add a param so you can check if its databse or tree
01:54:00rasherAt least until we move tree-talking to use the same method.. (still not sure it's a good idea though)
01:55:11 Quit barrywardell ()
01:55:27rasherJdGordon: thanks for helping me out
01:55:35JdGordonno worries :)
01:57:36rasherOdd, looks like it's already trying to talk, at least for files
01:57:49JdGordonwouldnt surprise me
01:57:50rasherLine 763+764
01:58:03JdGordonrun it in the sim and see what dir names its looking for
01:58:13JdGordonyou may not actually have to change any code
01:59:39rasherFailed to find: /Artist/
01:59:48 Quit spky ("byte me")
02:00:05rasher(while selecting "Artist")
02:00:38JdGordonso if you put your talk clips in the same "tree" as the tagnavi is setup as it shold work
02:00:56amiconnmarkun: There is another question regarding such an api. How much functionality does it need to implement by itself to be a true intermediate layer?
02:01:06 Part toffe82_
02:01:15rasherJdGordon: this doesn't take illegal characters into account though
02:01:25amiconnJust passing function calls 1:1 would be odd...
02:02:00amiconnTo me this is a very grey area
02:02:16Nico_Pamiconn: do you know the pcm code ?
02:02:25markunto me too, but it seems to be the way everybody does it
02:02:40rasherNo effort seems to be done for speaking files in the database - only directories
02:03:25 Join midgey [0] (
02:03:27 Join barrywardell [0] (n=barrywar@
02:03:45amiconnAnd there is not only the plugin api, but also plugin library components
02:03:58amiconnI'd rather stay away from gpl incompatible stuf...
02:04:10Nico_PI've been looking quickly at oscilloscope.c and it doesn't do any of the peak calculation itself... everything is done in the pcm code
02:04:27amiconn...even though that means I can't port XSkat, even if I would very much like to have a Skat engine in rockbox
02:05:10amiconnNico_P: It just uses the same data source as the peakmeter, and shows it in a different way
02:05:26ddaltonrasher: im on cygwin and my drive is at l:\ for my player and I am using SAPI5. So what should I type to run your script and make the .talk clips for the db?
02:05:44amiconnFor swcodec it uses the pcm peak calculation, for hwodec it queries the mas' quasi-peak registers
02:05:46Nico_Pamiconn: yes. I'd like to do something a bit fancier
02:06:07ddaltonsdoyon: I am half way through voicing the graphical eq. Have you already done this?
02:06:49Nico_Pamiconn: do you have an idea what data would be needed to perform a fourier/whatever transform to get a spectrum ?
02:07:05Nico_Pie is pcm_play_dma_get_peak_buffer enough ?
02:07:15amiconnSwcodec - the pcm data. Hwcodec - impossible
02:07:31rasherddalton: I don't know what config options the sapi engine expects. You'd have to check the configure script (or a Makefile created with it for generating voicefiles with sapi)
02:07:53Nico_Pamiconn: I'm doubting this kind of feature would e even wanted on hwcodec anyway ;)
02:08:08amiconnAnd I think that an fft is too demanding for our cpu, perhaps except the one overkill target
02:08:27Nico_Pthat's the one I'm thinking of... and we have asm ffts IIUC
02:08:36amiconnDo we?
02:08:49Nico_PI think someone told me that
02:09:00amiconnI only know about i(m)dct being in asm
02:09:21Nico_Pah, I probably got confused
02:10:39Nico_Pto get a spectrum from PCM data, what's needed is a DCT ?
02:11:00amiconnI'm not an expert at all in this area
02:11:17Nico_PI'm probably even less of an expert... ;)
02:11:44ddaltonsince pondlife's commit can I still use talk_menus_enabled ?
02:11:57 Quit midgey ()
02:12:44Nico_Pamiconn: do you know what kind of PCM data block would be needed ? and how to get it ?
02:13:08 Quit Domonoky (Read error: 104 (Connection reset by peer))
02:14:24kkurbjunJdGordon: Are you around?
02:14:58kkurbjundo you have any clues on why the screen update is so strange on the mrobe?
02:15:04 Join JavaMan22 [0] (
02:15:13JdGordonno, im going to try investigaing today
02:15:28JdGordonI thought it might be scrolling lines, but i dont tihnk so anymore
02:15:44kkurbjunyou hacked in something to make it partially work right?
02:16:05JdGordoni changed update_rectangle to just call lcd_update()
02:16:18JdGordonthat got the file browser going, but settings still donot show text
02:16:46JdGordonI asked jhMikeS and he though it might be FRAME getting corrupted or something
02:17:34kkurbjunI don't think it's FRAME
02:17:54kkurbjunI would think the screen would completely fail if that was
02:18:33JdGordonthe wierdest one is the running time screen partially shows text
02:18:50JdGordonit shows the 2 headers, but not the 2 dynamic lines
02:19:29JdGordonunless the lines are changing too fast for the lcd?
02:19:53kkurbjunthere is a controller that handles the LCD updates
02:19:53JdGordonthe disk info debug screen works (completly static info), but the stacks screen doesnt (completly dynamic)
02:20:02rasherJdGordon: only 10 more lists to go before your addition cancels out
02:20:22kkurbjunas long as it's written to the right location it would show
02:20:33JdGordonrasher: 10? 6
02:20:39kkurbjunit's similar to the gigabeat screen update
02:20:42rasherJdGordon: Depends on the target, really
02:20:51JdGordonscratch that idea... dircahce and db screens show fine also
02:21:03rasherOn h300 it's 12
02:21:31JdGordonwell, it was added more for simplicities sake, not for bin shrinkage
02:23:19 Quit barrywardell ()
02:23:42rasherHrm.. looks like the list of tracks doesn't use dirbrowse() at all
02:23:48rasherIn Database
02:23:58JdGordonit dort of does
02:24:17JdGordondirbrowse it the main function for both browsers.. the files that are shown are handeld differently in it
02:24:41mrkiko... goodbye all: I sho8uld go...
02:25:06mrkikoregarding the problem we where discussing (H300 crash with voice): will be a bug report be filled by someone?
02:25:13ddaltonsdoyon: around?
02:25:34rasherJdGordon: When showing files rather than directories, it never calls any of the ft_talk_* functions. Only when showing directories
02:25:36sdoyonThere. Committed. Now that's a couple of patches off my chest. Hope I didn't break too many things.
02:25:39ddaltonmrkiko: yes just submitted one. Please comment if you can add anything else.
02:26:28ddaltonsdoyon: did you voice the graphical eq menu? or should I do that?
02:26:45sdoyonrasher: my recent commit cleaned up tree.c somewhat, it might be a bit clearer now.
02:26:55JdGordonrasher: from line 932 in tree.c
02:27:03JdGordonthats called when the selection changes
02:27:46sdoyonThere a crash with voice on h300? Actually I have a problem on x5 that I tracked down to a recent playback change...
02:28:28rashersdoyon: that does look cleaner
02:28:32ddaltonsdoyon: do have the same problem sometimes as F#8003 describes?
02:28:36pixelmasdoyon: I wanted to ask you if the build table is accessible to you - one time there were a few broken builds. Just curious
02:29:21 Join Mil [0] (n=msavic12@
02:30:07sdoyonpixelma: Well it could be better. Are there color codes in there or what? Actually I'd like to see it when broken to find out if I'd notice :-).
02:30:18 Quit mrkiko ("leaving")
02:31:00rashersdoyon: there are colour codes, but anything not a zero indicates a number of errors or warnings
02:31:15 Nick Mil is now known as milivoje (n=msavic12@
02:31:19rasherUnfortunately some build servers generate bogus warnings
02:32:21 Join Yeeze [0] (i=57a64690@gateway/web/cgi-irc/
02:32:22sdoyonUh I get the feeling I'm not looking at the right table then... where is that?
02:32:29 Join nave7693 [0] (
02:32:31milivojehello, i would like to ask you about menus in rockbox: is there a way that they can be changed without messing with source?
02:32:39rashermilivoje: no
02:32:39pixelmasdoyon: those numbers are all links, if you select them you'll get the compile log but there are quite a few targets
02:32:52nave7693hi anybody know if I can flash my American e200 with a OF of a different region?
02:33:16rasherpixelma: perhaps builds with no errors shouldn't have the 0, but simply be empty
02:33:23 Quit Yeeze (Client Quit)
02:33:30 Join Yeeze [0] (i=57a64690@gateway/web/cgi-irc/
02:33:32JdGordonnave7693: you can, but you'll lose radio if you flash to the eu versino
02:33:33rasherWould be faster to navigate, I imagine. And look cleaner too
02:33:42amiconnsdoyon: Actually on that page there are 2 large tables. First table shows the number of warnings and errors by build, 1st line is most recent
02:33:52milivojeok, rasher, is there somebody (or you) that can help me with it? what shoul i edit in the source then?
02:34:02nave7693JdGordon: I just need to flash it with the japanese version so I can get my song titles to show
02:34:13amiconnThe second table shows the changes in binary size per build, top is most recent as well
02:35:10rashernave7693: this is not really Rockbox related
02:35:45amiconnThe table cells are coloured in addition to showing the numbers. For the first table, warnings will turn the cell yellow, errors turn the cell red. Normal is green
02:35:50nave7693rasher: no... but I got my answer anyway.
02:36:12milivojecan somebody tell me how can i make a theme for the whole rockbox, not just wps?
02:36:13amiconnThe number in the cell shows number_of_warnings + 10 * number_of_errors
02:36:45kkurbjunjdgordon, what did you change to get files to show properly?
02:36:59rashermilivoje: you'd need to look at apps/root_menu.c, but it's no picnic
02:37:17 Join J3TC- [0] (
02:37:22JdGordonkkurbjun: add lcd_update(); return; to the top of lcd_update_rect()
02:37:37J3TC-Windows: Setup a cygwin Rockbox development environment OR Setup a coLinux development environment OR Setup a VMWare development environment.
02:37:40J3TC-Which is better?
02:37:51J3TC-Or simpler/easier to setup?
02:38:05amiconnFor the second table, the numbers indicate how much bytes the binary size increased (positive numbers, cell turns red), or decreased (negative numbers, cell turns green). No change or minimal change gets no special colour
02:38:12scorcheJ3TC-: likely vmware
02:38:24J3TC-Cool, thanks.
02:38:26sdoyonamiconn: well, now that I'm looking at the right page, it'll help :-).
02:38:35rasherJ3TC-: vmware is probably easiest to setup, but not as easy to work with as cygwin
02:39:26sdoyonamiconn: Table is rather wide, and shouldn't it show the svn rev for which it's building?
02:39:56amiconnIt only shows a time stamp
02:40:09 Quit Yeeze ("CGI:IRC")
02:40:31 Part milivoje
02:40:33J3TC-I'll check up both then
02:40:35kkurbjunJdGordon: strange, that doesn't work at all for me
02:40:39amiconnMaybe it could also show the svn rev; just notice that the builds are not necessarily in consecutive svn rev order
02:40:59nave7693J3TC-: I have compiled in cygwin before
02:41:03JdGordonkkurbjun: can you see folders or nothing?
02:41:04nave7693J3TC-: vmware is overkill
02:41:20amiconnSome commits won't trigger rebuilds, e.g. commits to the manual or to rbutil, as those don't change actual rockbox code
02:41:22kkurbjunnothing worked at all
02:41:32J3TC-What do you mean overkill?
02:41:46nave7693J3TC-: installing linux to just compile the firmware is overkill
02:41:57amiconnAlso, if you do several commits while a build is currently running, all those commits will be included in the next rbuild round
02:42:10rashernave7693: it's also so much faster it's not even funny.
02:42:27 Quit Nico_P (Remote closed the connection)
02:42:46J3TC-Ah :3
02:42:47nave7693rasher: sure. if you are compiling so often just use the current builds
02:42:53amiconnsdoyon: Build table is looking good so far regarding your last commits. First one caused an overall size increase, but the others decreased size again
02:42:53rasherAnd easier (download VM image, open, run and you're away)
02:43:03 Join donutman25 [0] (n=chatzill@
02:43:43J3TC-I just want to make builds with applied patches to them
02:43:45rashernave7693: if you're setting up a dev environment, I assume you're not going to just compile Rockbox without modifying anything, so the current builds are not interesting in that case
02:44:10JdGordonkkurbjun: hmm.. it waasnt that. i removed the lcd_updae() line and the broser still works
02:44:37nave7693J3TC-: alright if you want to do dev work that's a different story
02:44:43*amiconn thinks that JdGordon should get a new keyoar ;)
02:45:11rashernave7693: why would you setup a compiler if you don't want to do dev work?
02:45:13JdGordonyeah, probably :p
02:45:31JdGordonkkurbjun: credits.rock works, so it doesnt look like a scroll or text issue!?!
02:45:48rasherWrong window..
02:46:14sdoyonamiconn: Thanks.
02:46:47amiconnrasher: :q!
02:47:43JdGordonrasher: sdoyon just made voicing a bit eaiser for you... look in tree_voice_cb
02:47:44J3TC-nave: nah, I just want to apply patches and build
02:47:56ddaltoncan someone tell me what the four options you can change in the graphical eq are?
02:48:08J3TC-So vmware is too much for that or a better option for that?
02:49:06scorcheJ3TC-: no...vmware is fine if you want something easy to set up and fast (it just occupies a larger footprint on disk and takes a bit more resources)
02:49:42J3TC-Anything less than 500mb is fine :3
02:49:52J3TC-I'm planning on installing this on a laptop
02:50:17rasherI think you'll struggle to fit any of the options in 500mb, to be honest
02:50:23J3TC-Or I guess I can install it my other computer and just run it there and d/l the ~2mb build
02:50:33J3TC-Wow..really...that big, huh?
02:51:28scorcheif you are that constrained, i would go with cygwin
02:51:41ddaltonrasher: in the graphical eq, do you know what the four options are when you press up and down?
02:51:48J3TC-Well, I guess I can go with cygwin.
02:51:53J3TC-I mean how harder can it be? >_>
02:51:59ddaltonI can't seem to figure it out in the code. But I think I can voice the db stuff.
02:52:22rasherddalton: I already told you. There are five, and they're bands in the equalizer - the exact same that are in the "advanced equalizer settings" menu
02:52:41ddaltonok thanks. But it only goes up to four...
02:52:48ddaltondoesn't it in the code?
02:53:11rasherNo idea.
02:53:51ddaltonoh it starts at 0
02:53:59ddaltonnot 1
02:55:14rasherStill, what's the point in voicing the graphical eq? It works almost exactly the same way as the menu, with only very minor differences in keypresses. The only advantage to sighted users, I think, is that it's all visible on the screen at once.
02:56:21ddaltonwell I am just about done so might as well finish it. And it is quicker to get to isn't it from the wps?
02:56:29 Quit Gnu47 (Nick collision from services.)
02:56:37 Join Gnu47 [0] (
02:56:49sdoyonrasher: Would you mind repeating a bit where you're at, and is there anything I can do to help?
02:57:21 Quit jhMikeS (Nick collision from services.)
02:57:25JdGordonkkurbjun: have you flipped the lcd yet?
02:57:27 Join jhMikeS [0] (n=jethead7@rockbox/developer/jhMikeS)
02:58:44rashersdoyon: I'm basically figuring out what needs to be done for .talk clips to work in the database browser. It seems that currently it looks for directory clips in /menuentry/ I need to modify this a bit so it would look somewhere saner where I could put talk clips generated by a perl script.
02:59:14pixelmaddalton: in the graphical eq screen the maximum I would voice is the current option you are on, not the values because I think you need to hear the changes to the sound (just my thought when I read that, not knowing what you have planned)
03:00:28rashersdoyon: This I can handle, but the problem comes when file or directory names are compounds of different tags. It's hard to generate a clip for that based on the database files. I would have hoped I'd be able to voice each part separately, but those parts are not available in tree_voice_cb - only the filename
03:00:42 Quit Buschel ()
03:01:06JdGordonrasher: i dont tihnk you need to voice each bit of the db
03:01:13JdGordonits the same as going down a dir tree...
03:01:18JdGordonyou know where you came from
03:01:21ddaltonpixelma: ok well im still trying to figure out how to voice those options and make it keep up with the voice.
03:01:22J3TC-Ok, I'm getting cygwin
03:01:43ddaltonpixelma: do you know if there is peek 1? or is there just peek filter?
03:01:48rasherJdGordon: but how about this: Playing file thumbnail: //02. Spacelab -
03:01:59sdoyonrasher: compounds of multiple tags? How is that?
03:02:05rasherJdGordon: I'd only have a talk clip for "Spacelab" in that case
03:02:28pixelmaoff to sleep now though, night
03:02:30 Quit mirak ("Ex-Chat")
03:02:33rashersdoyon: for example the filename is currently made up of track number, track name and then duration
03:02:41rasherAt least in some parts of the tree
03:02:58 Join webguest36 [0] (i=9db6c5cc@gateway/web/cgi-irc/
03:03:14rasherThis is difficult to create with pre-made talk clips
03:03:20 Part pixelma
03:03:34rasherI'd pretty much need a full tagnavi.config parser, and I'd love to avoid that
03:03:59ddaltonJdGordon: re your list change last night what needs to be done to put the info screen into a list?
03:03:59 Quit webguest36 (Client Quit)
03:04:13 Join webguest01 [0] (i=9db6c5cc@gateway/web/cgi-irc/
03:04:16JdGordonddalton: copy exactly how the debug menus using it do it
03:04:27 Quit webguest01 (Client Quit)
03:04:38 Join webguest01 [0] (i=9db6c5cc@gateway/web/cgi-irc/
03:05:29sdoyonrasher: You'll also need something that my other patches need: ability to queue up multiple thumbnails.
03:05:54sdoyonrasher: Tell me more about these filenames made up of multiple tags: are there other cases?
03:05:54 Quit webguest01 (Client Quit)
03:06:02rashersdoyon: also true
03:06:24JdGordonsdoyon: I dont need to manually call the get talk function right? its called by the list when the selection changes?
03:06:39ddaltonis r15235 the latest rev?
03:07:01rashersdoyon: I actually don't know much about the tagnavi syntax, but I imagine directory names could also be made up of multiple tags. For example an albumname with the release year added to the end.
03:07:50sdoyonJdGordon: Right, except you need to call gui_synclist_speak_item(), usually in the same place as gui_synclist_draw() gets called, such as on entering the list.
03:08:47sdoyonUhm... I guess I was naive, but what is tagnavi syntax?
03:09:01JdGordonits horrible :D
03:09:05rashersdoyon: it's the config file used for generating the Database menu tree.
03:09:54sdoyonrasher: I haven't mde more than short visits into the database as you can probably tell. Where's that config file?
03:10:06rashersdoyon: I think the tree_voice_cb in the case where it's called from the Database instead of trying to play the talk clip using the same mechanism, should ask the Database somehow which "parts" the currently selected item is made up of, and voice them
03:10:27rashersdoyon: it's in .rockbox/tagnavi.config
03:10:41ddaltonJdGordon: so what do I copy?
03:10:41rasherI don't really know how it works either - I'm simply using the default
03:10:50ddaltonfrom line 738 to 749?
03:11:08 Part nave7693
03:11:13JdGordonyes, and its callbacks
03:11:20 Quit tedrock (Client Quit)
03:11:41ddaltonJdGordon: can you tell me what line one of the call backs is on?
03:11:54JdGordonno, busy atm
03:12:28ddaltonJdGordon: ok so what am I looking for? a function variable?
03:12:29rashersdoyon: It's either that, or I need to implement the whole thing in the clip generation script, and I really don't think that's the simpler solution, to be honest. Especially since some of the values such as playcount and rating are dynamic (and should be voiced using the function to speak numbers)
03:13:52***Saving seen data "./dancer.seen"
03:14:36sdoyonrasher: agreed.
03:14:47ddaltonJdGordon: something like this?
03:14:49ddaltonstatic int dbg_threads_action_callback(int action, struct gui_synclist *lists)
03:16:20ddaltonJdGordon: so do I need a function like that for each item?
03:16:27 Join tedrock [0] (
03:17:14sdoyonrasher: Well looked only for 2mins so far, but it would appear that those tags get put together during db generation, and what we get out of the db are plain strings...?
03:17:49ddaltonsdoyon: did you get my email with the h300 keymap?
03:18:14rashersdoyon: The db generation puts tags into the database - the final strings are generated when each virtual directory is opened, I believe
03:18:31rashersdoyon: beyond that, I don't know much else
03:18:56sdoyonddalton: I did.
03:19:09ddaltonsdoyon: ok good is it ok?
03:20:04sdoyonddalton: haven't looked yet. You tell me :-).
03:20:39ddaltonsdoyon: your the programmer not me :-)
03:26:05ddaltonsdoyon: you said in a ml a while a go that you were working on speeding up playback. could you send me your work so I can take a look?
03:26:29 Quit Mouser_X (Read error: 110 (Connection timed out))
03:27:42ddaltonjhMikeS: around?
03:27:49JdGordoni hate to say it.. but the simplelist api works with the voicing now, so if you really want to you can make the debug screns voicable
03:29:15ddaltonsdoyon: what function should I use to play a beep in misc.c?
03:31:33 Quit spiorf (Remote closed the connection)
03:31:35 Join arkestra [0] (
03:31:55arkestragood evening everyone
03:33:38arkestramy my −− wuit tonight
03:34:21JdGordonnot quiet, just anti-social!
03:34:31arkestraah i c
03:34:48arkestrathen wa\hat are you doing in n IRC room/
03:35:28ddaltonsdoyon: can you help me with this info screen? or are you busy?
03:37:13arkestraa few quick ?s for the oldies here −− how is development going on the zune port, and need any help ?
03:39:18sdoyonddalton: Busy sorry. Reading up on tagtree.
03:40:38ddaltonsdoyon: ok
03:41:07ddaltonsdoyon: could you just tell me how to play a beep?
03:42:19sdoyonddalton: pcmbuf_beep()
03:42:42ddaltonsdoyon: does it take any arguments?
03:43:26sdoyonddalton: Look it up.
03:45:35ddaltonsdoyon: it has void pcmbuf_beep(unsigned int frequency, size_t duration, int amplitude); so will this do? pcmbuf_beep (22, 100, 10);
03:52:12 Quit criznach ("ChatZilla [Firefox]")
03:52:29rasherddalton: check other places that use that function. 22 is not a good frequency
03:52:59rashernor is 10 a good amplitude
03:54:37ddaltoncould i write pcmbuf_beep(5000, 100, 2500*5);
03:54:47ddaltonit is from playback.c and the original function thing is
03:54:48ddalton pcmbuf_beep(5000, 100, 2500*global_settings.beep);
03:55:05rasherWhy don't you use global_settings.beep as well?
03:55:21ddaltonbecause I want it to beep if beep is off as well.
03:56:16sdoyonThe beep setting is currently under playback options...
03:56:35sdoyonBut the beep setting is 0,1,2,3, why would you use 5?
03:57:19ddaltonoh ok it only goes up to 3
03:59:06ddaltonsdoyon: can you send me your updated set of patches?
03:59:10ddaltononce your done
04:00:08sdoyonddalton: Sure.
04:00:45ddaltonsdoyon: thanks. Can you send your speeding playback up one as well?
04:03:15sdoyonddalton: No. That one still needs adjustment, it's still rather rough, and I found a track that breaks it today. I'll put it up on the tracker when I can. Hard to choose between so many things to work on...
04:03:23 Quit SirFunk (Read error: 110 (Connection timed out))
04:03:43ddaltonsdoyon: ok
04:04:13 Join SirFunk [0] (
04:08:50ddaltonsdoyon: do you find the charger inserted and removed message annoying in my patch?
04:08:59ddaltonor your patch :-)
04:10:29sdoyonddalton: Possibly, as I said on the tracker. Although I can't say I've used it much. I imagine that's why you're looking at beeps, as I suggested on the tracker?
04:11:00ddaltonsdoyon: yes just about to upload a patch. I will remove the charging stuff from that patch and add the beep one in a new one
04:11:09ddaltonthey aren't really that related.
04:11:31ddaltonjust got to test my last build.
04:12:00ddaltonsdoyon: BTW what were you developing with the beep functionality?
04:13:00sdoyonddalton: ??
04:13:22ddaltonyou said you worked on a linux hand held device or something
04:13:54*rasher has talking Database working
04:14:03JdGordonwd :)
04:14:24sdoyonrasher: !!
04:14:34 Quit JavaMan22 ("Leaving.")
04:14:37rasherHold your horses
04:14:45rasherIt's only for "simple" strings
04:14:58sdoyonrasher: How?
04:14:58rasherThe ones under album for example, won't work
04:15:02rashersince they include trackname
04:15:18rashersdoyon: pretty simple really, check the patch - it's only a few lines changed in tree.c
04:15:31rasherShould be easy to follow
04:15:42rasherFS #7984
04:16:07rasherAnd I'm not sure I could explain it properly
04:18:28sdoyonOK, I see.
04:18:30 Join mirak [0] (
04:18:43 Join emeraldd [0] (
04:18:56sdoyonDo we need to know which tag is what, or just to have the individual strings delimited?
04:19:28rashersdoyon: you mean to speak "compound" entries? Just having the strings delimited should work
04:19:57rashersdoyon: actually not quite, now that I think of it. We need to know if it's a numeric value, a time value or a string
04:20:28rasherstrings should be checked in the /talkclips/ dir, numbers and time should be voiced using the voicefile
04:20:47rasherAgain, this seems like quite a bit of work
04:20:58sdoyonrasher: Yes I already meant to parse numeric strings. What does a time string look like?
04:21:13sdoyonrasher: probably not all that much actually.
04:21:44rashersdoyon: I'm not sure, but I believe Rockbox has functions for outputting time - that should be used, so it says "4 minutes 2 seconds" rather than "four colon zero two"
04:22:51rasherI presume (and hope) that value is internally kept in seconds (or milliseconds), and only formatted at the very last stage when putting it in the dirbrowser list
04:23:02sdoyonrasher: it does, but they take a parameter in seconds or milliseconds, so we just need to parse that.
04:23:30rashersdoyon: My hope is that we'd be able to get it from the database part as seconds, rather than the formatted string
04:23:40rasherAnd when I say we - I mean you.
04:25:32ddaltonsdoyon: Thanks for your update to p7910
04:25:37sdoyonrasher: you think the db has it parsed? Could be I suppose. Which tags are time?
04:26:18sdoyonddalton: My pleasure.
04:27:09 Quit Gnu47 ("I'm off to stalk my Pah'Mach'kai")
04:28:42rashersdoyon: I just checked the wiki. The database does indeed keep playlength in milliseconds.
04:29:26 Quit miepchen^schlaf (Read error: 110 (Connection timed out))
04:29:29rasherHowever, the formatting strings used in tagnavi are Lm and Ls - time in minutes and time in seconds (the remainder)
04:29:41rasherSo Lm:Ls gives the duration in human readable form
04:29:49 Join miepchen^schlaf [0] (
04:29:59rasherStill, that could just be voiced as Lm "minutes" Ls "seconds"
04:32:25rasherWhich is somewhat nasty, I guess.
04:33:19rasherThere will also almost certainly always be a colon between them that shouldn't be voiced I guess
04:34:34sdoyonAFAICT the format substitution happens in format_str(), called from retrieve_entries(). Looks like that's prepared ahead of time though.
04:35:14sdoyontagtree_get_entry() calls etrieve_entries() if realid >= current_entry_count ... I wonder when that is...?
04:35:34rasherNo clue
04:35:52ddaltonsdoyon: do you think a single beep for plugging the charger and removing it is ok?
04:36:08ddaltonbecause most people know if they are plugging the charger in or removing it :-)
04:36:11rasherddalton: I don't see why it should beep if beeping is disabled?
04:36:29ddaltonrasher: because it has nothing to do with playback
04:36:49sdoyonOTOH it should depend on some other setting, perhaps talking menus for now.
04:36:57rasherThen I think that setting should be moved out of playback
04:37:34rasherI guess a "blind-user friendly beeps" option could be added
04:37:37sdoyonThat might make sense too. But it might end up used in a lot of unrelated places. I have a beep in the keyboard now.
04:38:24*ddalton wow its getting hot here
04:38:34sdoyonI had a patch to play a short sequence of beeps to form a little tune/song. Makes it more informative. Never really used it though.
04:39:02ddaltonhow would you play a double beep? just put the two functions after each other?
04:39:12ddaltonnot that im going to just wondering
04:39:38ddaltonI will make it depend on talking menus
04:39:51ddaltonbut what if I want it to beep for my charger not playback?
04:39:52rasherIt's make sense to use two beeps in increasing frequency when plugging the charger, and decreasing when unplugging
04:39:53sdoyonI'm not sure actually, it's been months since I looked at beep mixing.
04:41:35sdoyonIn general, beep tunes like this are nice because they are shorter to listen to than a voice msg, and you don't need to interpret speech. It's faster, and also less disturbing if the player is playing a book or podcast.
04:42:12sdoyonAn alternative would be brief .talk clips with sound effects in them instead of speech. I used to have one for LANG_WAIT that sounded good.
04:42:13 Quit kugel ("Benutzer ist abwesend.")
04:50:18ddaltonsdoyon: done. What do you think of my last two patches?
04:50:33ddaltonp7910 and p8006
04:52:42sdoyonrasher: I imagine there must be a good reason to prepare the strings ahead of time. Who did you say knows this tagtree stuff?
04:52:58rashersdoyon: Slasheri and no one else, I think
04:54:00ddaltonsdoyon: what doesn't talk yet in the rb core?
04:54:29scorcherasher: bad bus factor =/
04:55:01sdoyonrasher: Perhaps format_str can be hacked to add delimiters in the output string. Those will probably need to be taken out before trying to display the text. It leaves us with parsing numbers and time. I'll probably have to parse nubmers anyway at some point.
04:55:56 Nick JRoT|koken is now known as JRoT (
04:55:59sdoyonrasher: Otherwise we'd need to store appropriate talk ids along with the name. But we'll have a sequence of talk IDs and clip names. Not sure how to represent that efficiently.
04:56:17JRoTwh00t it's 5AM already sleep well everybody
04:56:28sdoyonddalton: Sorry I'm working on that tagtree thing for the moment.
04:56:39ddaltonsdoyon: ok
04:57:07J3TC-I get an error with getting setup.ini using
04:57:59rashersdoyon: That might work. Perhaps it could add a delimiter and a type-indicator
04:58:19rasherIt seems a bit hackish though
04:58:51sdoyonrasher: Yes. To both comments :-).
04:59:02 Join sounddude [0] (
04:59:20rasherWould be nicer if you could simply ask the database code "which strings were used to generate the item n in the tree?"
04:59:51 Quit mirak ("Ex-Chat")
05:03:26sdoyonrasher: Yes. I imagine we can, I'm just wondering why they have them processed in bulk then. How much of this is cached...
05:03:34 Join Hammer89 [0] (
05:05:02rashersdoyon: presumably because the list code expects to be given an array of strings, ready to display
05:05:18 Join Mouser_X [0] (n=someone@
05:05:46sdoyonrasher: No, the list is given a callback function, and that function returns the string for one particular entry given by number.
05:06:00 Join sneakums [0] (
05:06:55sneakumson my ipod 5.5g, with recent builds the scroll wheel doesn't seem to work until the post-boot dircache update has completed. other disk activity doesn't seem to affect it. is this a known problem?
05:07:09 Quit XavierGr ()
05:08:00rashersneakums: I don't think think so, no
05:08:35rasherDo you know when it broke?
05:09:08sdoyonrasher: Do you know / can you explain the interaction with dircache?
05:09:39sneakumsright about now i wish i'd fixed up my patch to put the git ID in the build version
05:10:01rashersdoyon: beyond the fact that I belive it puts itself as a sort of layer between the regular file access functions, I don't know anything about it, no
05:10:20rashersdoyon: and I especially don't know how dircache and the database works together
05:11:33sdoyonrasher: That makes two of us :-).
05:12:31 Quit keanu|away ("ChatZilla [Firefox]")
05:13:56***Saving seen data "./dancer.seen"
05:15:03J3TC-Anyone know how to get rockbox's setup.ini for cygwin?
05:15:04sdoyonrasher: I'm afraid I'll have to leave it at that for tonight. Got to go.
05:15:21 Quit bb (Nick collision from services.)
05:15:26 Join bb_ [0] (n=bb@unaffiliated/bb)
05:15:38rashersdoyon: okay, thanks for looking into it
05:16:36rasherJ3TC-: sounds like a mirror problem.. try using instead
05:16:50sdoyonGood night.
05:16:51 Quit sdoyon ("ircII EPIC4-2.4 -- Are we there yet?")
05:21:26J3TC-Still won't for some reason
05:22:12J3TC-I think my firewall is screwing up
05:22:16 Join criznach [0] (
05:23:12ddaltoncan someone commit a very small patch for me?
05:25:10rasherYou're not going to get anyone to say yes without saying what it is..
05:25:48ddaltonrasher: it is a patch that beeps when a charger is removed or plugged in.
05:26:09ddaltongood for telling if it is actually charging and there is power running to the player...
05:27:43rasherddalton: I don't think it should depend on talk_menu. Personally I'd rather see a couple more of those beeps-that-blind-users-want grouped together, and a new beep option for those.
05:30:02ddaltonrasher: there aren't enough times you would use a beep
05:30:07ddaltonfor that sort of ting
05:30:30rashersdoyon mentioned the vkeyboard for one
05:31:05ddaltonso if I write up a quick patch to do this will it be likely someone will except it?
05:31:20rasherI don't know
05:31:28rasherThat was just my opinion about it
05:31:46ddaltonrasher should that setting be under "voice"?
05:32:20rasherThat's what I'm thinking, yes, even though it's not actually voice, but it would belong with those settings
05:33:22rasherAgain, it's just my own take on it - and I don't know if anyone else agrees. Finally, I'm not going to commit it because I don't feel comfortable adding it without any other input
05:34:08ddaltonOk I will write one up and see what people think/say...
05:35:18ddaltonrasher: what would you call the setting?
05:37:24rasherNo idea
05:40:29 Join midgey [0] (
05:43:04 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
05:44:04 Quit Mouser_X (Read error: 104 (Connection reset by peer))
05:44:11 Part Hammer89
05:48:41ddaltonrasher: ok I will just upload it and see what people say about the name.
05:51:42 Quit ddalton (Read error: 104 (Connection reset by peer))
05:52:11 Join ddalton [0] (
05:52:19ddaltonrasher: any progress with the db?
05:52:40rasherNo, not working on it.
05:52:48ddaltonoh ok
05:53:13rasherThe simple strings are working, anything beyond that is beyond me as well
05:58:09ddaltonrasher: do you know how to make a setting be on by default?
05:58:30rasherI don't know anything about the settings code
05:58:41rasherAgain, check how the rest of the code does it
05:59:22ddaltonrasher: ok I think I know
05:59:29ddaltonhow to
06:02:02 Join safetydan [0] (n=safetyda@rockbox/developer/safetydan)
06:02:45 Join Mouser_X [0] (n=someone@
06:12:24 Join Reno [0] (i=4c10ea95@gateway/web/cgi-irc/
06:13:52RenoHi guy's it's Reno, i was talkking to a person called TMM about a week ago a just want to know if you guys can tell that person for me that i wasn't able to get the ipod classic data sheets....sorry
06:14:07RenoMe and my friend tried are best
06:15:20emeralddI just spotted an odd check in the bookmark/playlist code, is there anyway for playlist_get_resume_info to return -1? The only return I see in the function is explicitly 0
06:16:14RenoI have a question i created some backdrops in photoshop for my ipod and they work and all but when i turn off my ipod and turn it on again it goes back to the original image? why?
06:17:24rasherReno: Because you failed to read the manual:
06:18:17rasherI believe this limitation could and should be fixed though
06:18:38Renookay thanks but can you please just tell me this? Is it possible to load different backdrops on my ipod and keep therm that way even if i shut it off?
06:19:13Renoi'm gonna read the manual again anyways. thank you
06:19:26rasherWell, the link I pointed you at explains it.
06:19:48Renoand...i'm really sorry abot not being able to get you guy's the ipod classic data sheets.
06:20:12rasherIt's not like we had our hopes high
06:25:26 Join psycho_maniac [0] (
06:27:13Renoi have another question. is it possible to change the rockbox bootscreen picture?
06:27:36safetydanReno, compile your own build is about the only way
06:29:12psycho_maniacwhy is car audio mode under system?
06:29:24Renowhat do you mean compile?
06:29:49ddaltonReno: compile from source...
06:29:59ddaltonpsycho_maniac: why not?
06:30:13Renoi've never really understood what that means. i even read it here in this site explaining what it is and i still dont really understand it?
06:30:50psycho_maniacim just curious. i have an idea of a place it should be but i want to hear a reason why it was put in system.
06:31:14ddaltonwhere do you want it?
06:31:37psycho_maniacwell i went to change it and thought. it has to do woth play/pause so i figured it would be in playback.
06:31:53ddaltonReno: well then I doubt you will have much luck with compiling. Basically setup your environment and then run configure then compile
06:32:04ddaltonof course you need the source
06:32:22ddaltonpsycho_maniac: ok can you program?
06:32:56psycho_maniacnope. like i said i dont DEMAND this i want to know the reason why it was put in system.
06:33:27Reno2 questions: 1. is compiling hard? 2. what do you mean source?
06:33:28ddaltonok I don't know
06:33:36ddaltonbut move it if you don't like it :-)
06:33:49ddaltonthe source code.
06:34:09ddaltondepends if you can program or if you have done any of this before...
06:34:20ddaltonpeople who can't program can still compile
06:34:33psycho_maniaci guess theres only 3 people in here so i dont know if anymore people like this idea. its just a discussion.
06:34:38Renodo you you guy's have any toturial videos on youtube or something?
06:34:49safetydanReno, there's a guide to compiling on the wiki
06:34:50psycho_maniacjust look in the wiki
06:34:52ddaltonpsycho_maniac: how is a reason going to help? do you want me to write you a patch later?
06:35:45Renoi'm sorry but like i said i've read that wiki luck...i suck! =(
06:36:17psycho_maniacaparently you didnt because i read it with no clue what i was getting into and i now compile my own builds
06:36:21ddaltonReno: well what don't you understand
06:36:36ddaltonit isn't to complicated
06:37:54Mouser_XReno: When people write a software program, they (usually) write it using some form of text. The document that is created is refered to as the "source."
06:38:39Mouser_XWith a compiler, it can take that source, and "translate" (loose term. It's not exactly correct) it into an executable.
06:39:32Renouuummm....pretty much everything....can you guys give me a link to that guide again please so i can read it just one more time and see if i understand any of it this timer please?
06:39:36 Join homielowe [0] (
06:39:48Mouser_XReno: There's this, if you didn't read it.
06:40:02Renois there a way to just modify the rockbox that i already have?
06:40:08Mouser_XNo, there's not.
06:40:15Mouser_XYou have to modify the source code.
06:40:20Mouser_XWhich you don't have yet.
06:40:21Renook thanks anyway! =)
06:40:48ddaltonit compiles the source to an executable code that can run
06:41:04ddaltonso from human readable language to machine language...
06:41:10Renook. seriously maybe that is what i dont understand. where do you get this "source Code"?
06:41:27Mouser_XThat page I linked tells you how, and where, to get the source code.
06:41:42Mouser_X(How is the hard part.)
06:42:13psycho_maniacif you can double click on a exe, extract an 7zip file and do "/run command" commands. you can do it
06:42:33kkurbjunJdGordon: I do have a rotated screen option, but it's noticeably slower with data and code cache disabled and I havn't done the partial updates with lcd_update_rect.
06:43:05Mouser_Xpsycho_maniac, Reno: Perhaps, more specifically, setting it up to *do* the "how" is the hard part.
06:43:30Mouser_XThough, with the VMware image, it's much much easier.
06:43:46psycho_maniacyes. thats what i was refering to .
06:43:47Mouser_X(Though, I've never been able to get that to run in a useable way...)
06:44:04scorcheMouser_X: why not?
06:44:15emeralddgentoo it pretty easy to get working as well but there's a entirely different learning curve there
06:44:25Mouser_Xscorche: I don't know. It killed my PC when I tried to run it.
06:44:35scorcheMouser_X: define kill
06:44:39Mouser_XIt made it horribly horribly slow.
06:44:51Mouser_XJust running it, not compiling.
06:44:52scorchewhat are the specs of your computer?
06:44:55Renoi'm actually not on my computer right now do you guys know how i can download that guide website so i can use it on my laptop? it's got no internet.
06:45:29scorcheemeraldd: he is using windows...
06:45:36scorcheReno: just save the page to a file
06:45:37Mouser_Xscorche: My PC (not my laptop) is a 2 ghz AMD 64 bit CPU, with 512 MB of RAM (as I recall. It might be more RAM...)
06:46:05scorcheemeraldd: why go into that when he can just save the page through a submenu though?
06:46:16Galoisthat's not enough ram to usefully run vmware these days
06:46:23scorcheGalois: sure it is
06:47:23 Quit jhulst ("Konversation terminated!")
06:47:25scorcheMouser_X: did you lower the ram line in the vmx file?
06:47:34Galoismaybe if you use outdated host and guest OSes like redhat 9 or windows xp, then it is
06:47:36Galoisotherwise it's not
06:47:41emeralddscorche: I was thinking the page had child pages
06:47:50Mouser_XI'll say no (because I almost certainly didn't) because I don't know what a VMX file is.
06:47:51emeralddwget -np -m usually does the trick for me
06:48:30scorcheMouser_X: it is the vmware configuration file that is used with the image
06:48:55Renohow do i save the page to a file?
06:49:04Renoi'm using fire fox
06:49:08Mouser_XI recall someone saying to lower RAM usage in VMware, but they never said how.
06:49:20scorcheReno: right-click > save page as..
06:49:20 Quit linuxstb (Read error: 110 (Connection timed out))
06:49:30Mouser_XReno: Right-click on the page. What comes up?
06:49:37scorcheMouser_X: that is change the line to memsize = "256"
06:50:12Mouser_XWell, when I get my PC running again (I need a new power supply), I'll try that out, and see how it goes.
06:50:55scorcheReno: keep in mind that you will need to save an archive of the souce and cant use svn to get teh source, as your laptop lacks a network connection
06:51:55Renoi use windows vista on my laptop can i compile and stuff there?
06:52:50Mouser_XReno: Yes, I think all the software you need (be it VMware, or cygwin) should run under Vista (though, I'm not certain on that).
06:52:53Renowhen i right click it says save page as...
06:53:18Mouser_XSo use that option.
06:54:19Renoyeah but save what text files, webpage compleate, html only... wich?
06:56:10Renook never mind...
06:56:47Renoi took a skreenshot with screen grab, i should have thought of that earlier.
06:59:05 Quit midgey ()
07:00:39 Quit ToHellWithGA ("You know you'll miss me a lot.")
07:00:43ddaltonMouser_X: cygwin will. But you need atleast 2 gb ram under vista for the vm probably
07:00:46Renook i got somestuff now. what do i need to download to be able to compile and all that on my laptop?
07:00:53ddaltonor maybe 1.5 if they even make that
07:02:05Renooh, by the way thanks for helping me so much guys! =) guys busy right now?
07:05:42 Join XavierGr [0] (
07:10:16psycho_maniacwhat is the problem reno?
07:11:32Renocan you guys tell me what i need to download to be able to compile and all that other stuff on my ipod?
07:11:41Renoplease? :)
07:13:37psycho_maniacit says what to download in the wiki.
07:13:47Mouser_XReno: The amount of stuff you need to download can't be listed here. There's too much. That page I linked "SimpleGuideToCompiling" tells you what to do.
07:13:59***Saving seen data "./dancer.seen"
07:14:23 Quit ddalton ("leaving")
07:14:29Mouser_XOr, check out for a *slightly* easier method.
07:14:45Mouser_X(Though, there's still lots of stuff you'll need to download.)
07:15:04psycho_maniacreno: are you going to be using vmware or cygwin?
07:16:05Mouser_Xpsycho_maniac; Reno: Either way (cygwin or VMware) there's a lot of stuff you'll need to download. The source code is roughly 300-400 MB (at least, that's what it seems like on my laptop here).
07:17:02Renook what is the basic stuff i need
07:17:12psycho_maniaclook at the wiki and it will tell you
07:18:10Mouser_XReno: If you follow the directions in the wiki, you *can* download everything you'll need into 1 directory, and then put that directory into a (BIG) archive (such as a ZIP file). You'll need to follow the wiki to do this though. I can't tell you here. Again, there's too much. That's what the wiki is for.
07:18:20Renook i have to go now. THAK YOU GUY"S ALOT!!!!! THANKY YOU VERY VERY VEERRYY MUCH!!!! :)
07:18:41psycho_maniacok bye
07:18:56Renoaging thanks! :)
07:19:54Renoand if i am ever able to get the ipod classic data sheets (which i still wont quit trying to get them) i'll let you guys know! bye! :)
07:21:04 Quit Reno ("CGI:IRC (EOF)")
07:34:18 Part emeraldd
07:35:20 Join midkay [0] (n=midkay@rockbox/developer/midkay)
07:58:07 Quit sounddude ("Leaving")
08:07:13 Join Rob222241 [0] (
08:07:14 Quit criznach ("ChatZilla [Firefox]")
08:17:04 Quit feindbild (Remote closed the connection)
08:23:39amiconnJdGordon: A simplification that increases code size??
08:25:46 Quit Rob2222 (Read error: 110 (Connection timed out))
08:26:36JdGordonamiconn: the best type !
08:27:53amiconnIt should become smaller on SH1 and also probably on arm if you limit the number of parameters to simplelist_info_init() to 4
08:27:57*JdGordon doesnt know why that commit increased bin size though... i would have thought it would shrink it
08:28:02amiconn(right now there are 5)
08:28:11JdGordonok, ill remove one
08:28:33JdGordonany idea why he ondio fm size didnt change though?
08:28:49JdGordonor just becasue the previous build was larger than it should have been?
08:28:58amiconnBecause it is biased by the build server hopping
08:30:13amiconnThe callback data pointer is often NULL, and it would make sense to drop that and just init it to NULL afaics, since the callback function itself is also just inited to NULL by default
08:30:57JdGordoni was going to drop the selection size one which is used even less often than the callback data
08:31:18amiconnI wonder whether reverting that commit and using designated initializers in the calling functions would be even better...
08:32:02amiconnI mean, remove the init function, and use:
08:34:05amiconnstruct simplelist_info { [title] = "Database Info", [count] = 7, [selection_size] = 1, [database_callback] = database_callback, [hide_selection] = true } info;
08:34:21amiconnOf course brokenn into one line per value
08:34:36JdGordonwell.. that commit was really to simplify it more than to shrink bin size
08:35:00JdGordonback in 10min
08:36:13amiconnThe advantage of that form of initializer is that members without explicit assignment are automatically initialized to zero
08:37:09amiconnEh, sorry, the example wasn't correct for a struct. Should of course be:
08:37:55amiconnstruct simplelist_info { .title = "Database Info", .count = 7, .selection_size = 1, .database_callback = database_callback, .hide_selection = true } info
08:38:07*amiconn confused it with the array form
08:38:21JdGordonyeah, i thought the [] form looked wrong...
08:38:31JdGordonbut that way is still more work for the coder
08:38:52amiconnYOu don't need to list all the elements that should be NULL or 0
08:39:55JdGordonalso, 1) imho that looks more messy than explicilty initing every member, and 2) 0/NULL isnt a same default for all the members
08:43:04amiconnI didn't say it'd be a sane default. But several members are NULL / 0 in the places you converted.
08:43:50amiconnIt was just a thought to reduce the writing in the caller without having an init function
08:43:58JdGordonamiconn: changing topics... on the mrobe some lists work fine, but some everything is displayed except the item texts (tilte, statusbar, icons and scroll bar all show). any idea what function could be broken to cause that?
08:44:12JdGordonive rmeoved one param, we'll see what the delat is
08:46:52amiconnThe m:robe uises standard lcd-16bit.c?
08:47:17JdGordonand there doesnt appear to be anything wrong with lcd_update() and lcd_update_rect()
08:49:36amiconnAnd text output works in other places?
08:49:43 Join pixelma [0] (i=pixelma@rockbox/staff/pixelma)
08:49:59JdGordonyeah, no problem displaying text in the menus and in credits.rock
08:50:11JdGordonbut in the setting screens the text isnt shown
08:50:35JdGordonand in the fonts/config/wps/etc browsers no text, but icons are ok
08:50:42JdGordonand its fine in the regular file/db browser
08:51:24amiconnThese are the very same browsers...
08:51:52JdGordoni know...
08:51:56JdGordonit doesnt make sense to me
08:51:58amiconnWhat kind of lcd hookup does the m:robe have?
08:52:22JdGordonumm... i dont know
08:53:18amiconnWell, you said that lcd_update[_rect]() is ok, so how would you know that?
08:53:32JdGordonbecause calling it displays the change fine
08:54:53amiconnLCD is memory mapped like gigabeat and e200 afaics
08:55:18amiconnSo I'm almost sure it's a caching issue in lcd_update[_rect]()
08:56:01JdGordonok, so any suggestions on how to try and fix it?
08:56:14kkurbjunamiconn, I believe that data and instruction cache are disabled right now
08:56:21kkurbjunon the mrobe
08:56:30amiconnBelieven or know?
08:58:36kkurbjunthe loader should be disabling it.. I havn't reviewed the asm in detail, but the comments indicate that that is what is the intended result
09:00:22kkurbjunthis is from cat/shirour so I didn't write it myself..
09:00:24 Join stevenm [0] (n=stevenm@
09:01:56stevenmnls: n1s: for the log.. you mentioned the midi pitch bend depth causing a ringing artifact. I can try to have a look... could you send me a copy of the test file please?
09:02:01 Quit stevenm (Client Quit)
09:08:02kkurbjunJdGordon: do you know where the file drawing code takes place? I am still not seeing a proper file view
09:08:17JdGordonin gui/list.c
09:08:44JdGordonsorry.. that actually calls text_area_update() which calls lcd_update_rectangle()
09:10:36 Quit Mouser_X (Read error: 110 (Connection timed out))
09:10:36 Quit Rick (Read error: 104 (Connection reset by peer))
09:11:02 Join Rick [0] (
09:11:19JdGordonamiconn: say it is a cache problem, how do we go about making sure?
09:12:21 Join ToHellWithGA [0] (
09:12:21 Quit Rick (Read error: 104 (Connection reset by peer))
09:13:04 Join Rick [0] (
09:14:03***Saving seen data "./dancer.seen"
09:17:59 Quit psycho_maniac (" HydraIRC -> <- *I* use it, so it must be good!")
09:20:03 Join stevenm [0] (
09:21:13stevenmHello. I am trying to debug a memory corruption within a plugin using GDB. I have a pointer that is being overwritten somewhere. Is there a way to determine where? I tried to just 'watch' that structure but it didn't do anything. Am I missing something?
09:23:30JdGordonkkurbjun: all those lists and browsers are drawn by the list which eventaully calls gui_textarea_update() which calls display->update_rect(0, y_start, display->width, y_end - y_start);
09:23:39JdGordonthere is no reason why some should work and some shouldnt?
09:24:49kkurbjunso they are all full width updates..
09:25:05 Quit ToHellWithGA (Remote closed the connection)
09:25:31kkurbjunI'm really not sure, that assemble that's in the loader should be disabling the data cache
09:25:41kkurbjunand the MMU is disabled
09:26:23kkurbjunI'm looking through it now
09:26:37kkurbjunand it looks correct from what I can see
09:26:44JdGordoncan he FRAME be moved to somewhere which definatly isnt cached?
09:26:48kkurbjunthe instruction cache is enabled
09:28:00kkurbjunyou could move it wherever, but the MMU on this system determines if different addresses are cached or not from my understanding, and it is disabled..
09:29:13 Join Crash91 [0] (i=d98bd903@gateway/web/cgi-irc/
09:29:43Crash91hi guys
09:29:56Crash91my friend wants his ipod to get rockboxed, but he still wants to use itunes
09:29:59Crash91is it possible?
09:30:04stevenmThis makes no sense. Is there something different about watchpoints in plugins than in regular C programs?
09:30:29stevenmA variable is being blatantly changed. Legitimately. And GDB just continues on even with a HW watch
09:31:18Crash91JdGordon: Thanks, :)
09:31:36JdGordonkkurbjun: there was no corruption in the credits plugin which does lots of lcd writing... so...
09:31:40Crash91you have to be admin to run ipopatcher right?
09:31:52 Quit XavierGr (Nick collision from services.)
09:31:55 Join XavierGr [0] (
09:32:03kkurbjunJdGordon: looks like the data cache can't be enabled without the mmu from the arm programmers refrence
09:36:53JdGordonso that means its not a cache issue?
09:37:24JdGordonor the mmu is being enabled accidnently?
09:39:14kkurbjunI don't think it's being accidentally enabled, there are very specific commands to enable both
09:39:34kkurbjunyou said that lcd_update_rect is working correct right?
09:39:48kkurbjundid you actually test partial width writes?
09:39:54JdGordonwell.. it seems to be working
09:40:14JdGordonno, ill give it a proper test
09:41:17kkurbjunDoes anyone see any glaring problems with this code?:
09:41:56kkurbjunwow, it's late, I need to sleep
09:42:08JdGordonhehe goto bed then :)
09:42:16kkurbjun:), night
09:43:43 Quit Crash91 ("CGI:IRC")
09:43:56 Join ddalton [0] (
09:46:55amiconnkkurbjun, JdGordon: All I can imagine atm *is* cache problems - and they would show up more often if the screen in question does *not* do lots of updates, but rather just a few ones
09:49:26JdGordonamiconn: the menu screen refreshes every HZ ticks, the setyting screen blocks...
09:50:01ddaltonhow do I get the diff since the last applied patch?
09:50:26ddaltonI have a patch that depends on another one.
09:51:43 Join bluebrother [0] (i=ojM367Rd@rockbox/staff/bluebrother)
09:52:40JdGordonamiconn: other than the timeout on the get_action() call the menus and setting screens redrwa the same amount, and I just changed the setting screen to timeout after HZ so it is the same but no change
09:58:55Slasherii am just fixing the simplelist code so that it can refresh the screen automatically
09:59:09JdGordonit needs to?
09:59:17Slasherithat is necessary to make the tagcache debug screen show the last accessed file synchronously with tagcache
09:59:47Slasheriit only does auto refresh if callback returns ACTION_REDRAW
09:59:54ddaltonamiconn: how do I turn off rec but stay in the rec screen? I want to make it so when I press a certain button this happens.
09:59:55Slasheriso it doesn't do that always then
10:04:10 Join Thundercloud [0] (
10:05:17 Quit homielowe (Remote closed the connection)
10:06:16stevenmstill noone knows why watchpoints don't work in plugins?
10:07:31 Join ompaul [0] (n=ompaul@gnewsense/friend/ompaul)
10:07:37*JdGordon thinks his mrobe has overheated!
10:08:47ddaltonJdGordon: Do you have time to help me with this list stuff?
10:11:27JdGordonddalton: its not that difficult.. just look at the examples of where its being used and figure it out
10:12:30ddaltonjdGordon: ok what does info.selection_size = 2; mean?
10:23:22 Join bertrik [0] (
10:23:38stevenmhey amiconn, are you around?
10:26:18amiconnddalton: I already answered that question: you cannot. I also explained why.
10:26:36amiconnstevenm: yes, sortof
10:26:56stevenmamiconn, I hear you know how to use GDB and am in need of help.
10:27:13stevenmamiconn, is there any special procedure to using 'watch' on the Simulator?
10:28:00stevenmamiconn, I gdb rockboxui, set a break within my plugin, then watch my variable. But the watchpoint is never tripped
10:29:26stevenmamiconn, do you have any idea why this could be?
10:30:02amiconnI have no idea. I sometimes use gdb, but when I do, it feels cumbersome
10:30:19amiconnI never used watchpoints in gdb yet
10:30:40stevenmamiconn, well, I have something overwriting a variable that shouldn't be overwritten. How would you go about debugging this?
10:30:43amiconnBreakpoints do work, also in plugins, if you let it load the symbols
10:31:22stevenmamiconn, at one point the pointer points to legitimate memory, then it points somewhere way out to space. Nothing every writes it besides the initial allocation.
10:31:48amiconnIf all else fails, I would do this on a coldfire target, with the 'catch mem accesses' feature hand-tweaked to trigger on the access
10:32:20stevenmamiconn, what do you mean? how can I do that?
10:33:25stevenmamiconn, does it take in a pointer? I don't even know if the variable being overwritten on the sim is the same one as on the target
10:34:04 Part pixelma
10:34:13stevenmHAHAHAHA. This is new. When I try to select 'Catch Mem Accesses' on H300, it locks up when I hilight View IO Ports
10:35:12amiconnThe coldfire has some realtime debug support. It can trigger a special interrupt based on access masks (address, address mask bits and access mask bits)
10:35:27stevenmamiconn, yeah........ once in the Debug menu, pressing Up or Down scrolls the cursor once and then locks up the device. Tried resetting it. This is the fifth time
10:36:25amiconnThe 'Catch mem accesses' feature uses this to trigger on accesses to 0x00........ on coldfire, which e.g. triggers on NULL pointer accesses (also with an offset, e.g. accessing struct members)
10:37:35stevenmHmm. I suppose it would work. GDB seems like it should be working, but isn't.
10:37:38amiconnOnce I had a similar problem as you, and couldn't find it with conventional methods
10:37:44 Join linuxstb_ [0] (
10:37:48 Nick linuxstb_ is now known as linuxstb (
10:37:54stevenmI can't physically get to the Catch Mem Accesses option though.
10:38:05stevenmThe UI locks up the second I enter the Debug menu
10:38:15stevenm(on the target)
10:38:15ddaltonamiconn: what files would need to be rewritten so we could play a beep in recording?
10:38:20Slasherion H100, it immediately panics when i try to scroll the debug menu
10:38:26amiconnThen I changed the programming of that feature to trigger on the overwritten variable instead of the zero area
10:38:41Slasherimaybe it has something to do with logf, already fixed a few bugs with that
10:38:47Slasheritrying on simulator now
10:38:54amiconnSlasheri: Then the debug menu does something nasty
10:39:28amiconnBasically, if no uninitialized pointer accesses or similar nasty stuff happens, enabling this feature shouldn't be noticeable
10:39:48amiconnstevenm: ...and I found it. An array access was ot of bounds....
10:40:35amiconnHaving e.g. int a[4]; .... a[4] = lol; is a bad idea, isn't it?
10:40:53Slasheriwhere did you find that?
10:41:07amiconnIn some plugin, months ago
10:41:25amiconnAlso, this problem only manifested on archos, although it was wrong for all targets
10:41:40amiconnArchos crashed, probably because of the order of variables in memory
10:42:01amiconnOn SH1 we also have that 'Catch mem accesses' feature, but not on arm
10:42:25Slasherithere was also some wierd stuff with cpu boost logging, two methods are called that should return a pointer to a str but actually doesn't have any return in them
10:42:56amiconnSTrange... no problems with 'Catch mem accesses' enabled on my X5 the debug menu
10:43:15Slasheriok, it's probably a logf related then
10:43:18amiconnBut I don't have the latest revision installed
10:43:36Slasherieven without catch mem accesses, there is immediately a kernel panic
10:45:59 Join Xerion [0] (
10:46:38amiconnjhMikeS: around?
10:47:20amiconnThere is a problem on my H10 that seems to be related to your thread locking stuff
10:47:50amiconnI have my start screen set to the dir browser
10:48:18amiconnIf I have dircache enabled, *no* button works in the browser while the dircache scan is running in the background
10:48:30amiconnThe buttons don't even pre-register
10:48:49amiconnThis doesn't affect PP5022 - tested on iPod video
10:48:55jhMikeSNo spinlocks are kept locked outside of task switching
10:49:22 Join kubiix [0] (
10:49:23amiconnThis behaviour is new..
10:49:29 Quit stevenm ("Connection reset by beer")
10:49:43jhMikeShmmm. I did have a similar experience though. I also noticed something similar on e200 but much earlier than with DC.
10:50:31jhMikeShmmm...maybe I left out an interrupt reenable somewhere
10:50:49*amiconn builds for mini G2 just to check another PP target
10:50:58amiconnPP5002 (2nd gen) is also unaffected are the coldfire targets
10:51:25 Join davina [0] (
10:51:33jhMikeSthat would be built like any other single core.
10:52:29jhMikeSyou know what? I think I may know what it is already
10:52:53jhMikeSI did leave out an interrupt enable...want to try a one-liner?
10:53:21 Join hannesd [0] (
10:54:02jhMikeShold, on I have modded thread.c here...I need to get the line number on SVN
10:55:54jhMikeSthread.c line 2390 - add set_irq_level(oldlevel) just before the return
10:56:12jhMikeS#if NUM_CORES > 1
10:56:41jhMikeSin any case it should be there :)
11:01:20amiconn(will take a while)
11:01:32jhMikeStakes that long just to make bin?
11:02:00amiconnI am doing a full rebuild to make sure.
11:02:32amiconnA different question: How would you proceed if you need to find the lcd init sequence for the H10 in rom or OF.bin?
11:03:04jhMikeSI think I did already...would need to find it again
11:03:21amiconnIt looks like the lcd driver uses structs all over; I was able to find some places where this struct is initialized, but not where the functions are which actually use it
11:03:33jhMikeSsomewhere searching for lcd-related regs got me there (or parts of addresses)
11:03:57 Quit bertrik ("booting to linux")
11:03:58amiconnI need the lcd register parameters for the small h10 so I can implement lcd on/off etc
11:04:32amiconnI already have a datasheet for that at lists the registers for TL1771 and their relevant bits
11:04:41amiconns/at/at least/
11:05:32amiconn..and I also know the correct R_ENTRY_MODE bits so I could implement asm'ized lcd_yuv_blit() right away
11:06:09amiconnIdeally we can also find where 0x70008a00 and 0x70008a04 are written, in order to find out what the bits mean
11:06:19jhMikeSASM optimizing week? there's been a flurry of codec/lcd optimizing recently.
11:06:43amiconnSetup for small H10, big H10, Nano and Color in these regs is all different
11:07:15amiconnsmall H10 and Nano run the block controller at nearly optimal speed when boosted (but could go faster when not boosted)
11:07:16jhMikeSI did stuff like searching all instances of 0x70000000 throughout with a simple objdump
11:07:27amiconnColor runs it too slow - could go much faster
11:08:03amiconnjhMikeS: I found several places where 0x70008a00 is put into a struct - but I have no idea how to find out where this struct is used
11:08:31amiconn...since the functions which write to this struct get the address passed from somewhere else
11:08:32jhMikeScould be dynamically allocated
11:08:57amiconnSo how could I figure this out?
11:09:21amiconnI guess an emulator would help - but the existing emu is for sansa, and I have no MSVC to poke on it
11:09:30jhMikeSI can't find a formula except to find code that looks like it writes to such a struct.
11:09:40amiconn...and no experience at all with emulator coding
11:09:42jhMikeSyou can download MSVC
11:10:04jhMikeSThe basic one is free and compiles the emu
11:10:07amiconnYes, the 2005 express version - which reportedly doesn't work for compiling the emu
11:10:16amiconnAt least the fs task says so
11:10:19jhMikeSWell, I'm using it must :)
11:10:31jhMikeSYou also need the Platform SDK
11:10:34 Join bertrik [0] (
11:10:46jhMikeSdownload that and tell it where the libs are and it's good to go
11:11:07*amiconn might need to also free some space on his hdd :/
11:13:18 Join mf0102 [0] (n=michi@
11:13:20amiconnDid you try to use this emu for h10?
11:13:26 Join SkinInd95 [0] (
11:14:06***Saving seen data "./dancer.seen"
11:14:08 Quit SkinInd95 (Client Quit)
11:14:33jhMikeSnot yet...needs some work to do it
11:15:05jhMikeSthough it has been used for c200 so I don't know if that helps at all
11:15:17amiconnUnfortunately it doesn't
11:15:29jhMikeSmy version has fixed threading as well so it shuts down properly
11:15:53jhMikeSit will otherwise won't terminate at all
11:16:14amiconnc200 is flash based too, i.e. ata isn't needed, and the lcd is hooked to the mono bridge, not the colour bridge
11:16:37jhMikeSmono bridge? I thought it was a color target.
11:16:47amiconnYes it is, yet it uses the mono bridge
11:17:07amiconn(the one at 0x70003000 which is much simpler and slower than the one at 0x70008a00)
11:17:17 Join ender` [0] (
11:17:45amiconnThat's the reason why it maxes out at just 38fps when boosted, even with that tiny resolution
11:17:51ddaltonjhMikeS: what code is executed while recording?
11:18:06jhMikeSddalton: that's one huge list of code!
11:18:07amiconnjhMikeS: YOur fix is working :)
11:18:32jhMikeSamiconn: I though it might be that. interrupt would've stayed disabled until the next processor sleep.
11:19:00amiconnShall I commit? (since you said you have other changes in thread.c)?
11:19:09jhMikeSdoesn't matter. go ahead.
11:19:13ddaltonjhMikeS: ok so is there a function I can just call to run that while recording?
11:19:33Slasherihmmh, the debug menu is completely broken.. doesn't work even without logf :/
11:19:50amiconnSlasheri: huh?
11:19:54amiconnIt works here...
11:19:57Slasheriit just crashes..
11:20:11*jhMikeS has been using it consistently
11:20:25amiconnSlasheri: Oh, hmm, latest svn does crash it
11:20:39jhMikeSddalton: not sure what you mean.
11:20:42ddaltonmy settings aren't been saved on a shut down of my h300.
11:20:47ddaltonanyone know why?
11:20:49amiconnEither it's JdGordon's simple list stuff, or sdoyon's voice callbacks :|
11:21:13*ddalton Hates debugging
11:21:51ddaltonjhMikeS: Well is there a function that is called while recording to save the recording file to disk and what ever else?
11:21:55*amiconn can't think of a good commit message for that threading fix...
11:22:23ddaltonI want to make a quick memo feature. While I hold rec down record but once I let it go just return to what I was doing before.
11:22:26jhMikeSddalton: that's in pcm_record.c
11:22:30ddaltonand this should work from anywhere.
11:23:02Slasheriah, it's that simplelist stuff
11:23:06ddaltonjhMikeS: what should I look for there?
11:23:08jhMikeSddalton: audio_stop_recording does flushing
11:23:12JdGordonSlasheri: which debug menu?
11:23:20Slasherithe main debug screen
11:23:26Slasheriwhen scrolling, it crashes immediately
11:23:50JdGordonwtf? thats a bit odd
11:23:58jhMikeSddalton: I'm not sure why those functions need using outside the recording screen or a recording plugin.
11:24:07amiconnddalton: This cannot work from anywhere. Recording needs lots of stuff set up differently than playback
11:24:07Slasheriat least it's missing the simplelist_info_init
11:24:19Slasheriand initializing directly those parameters, not sure yet if that causes it
11:24:54JdGordonhmm woops.. missed it, but not calling init sholdnt cause it to crasg
11:25:12ddaltonjhMikeS: so could I write something like: while(button ==action_std_rec) {what ever code needs to be executed when recording. if(button !=action_std_rec) {code to kill recording and save.}}?
11:25:56ddaltonamiconn: im going to enter the rec screen and start rec automatically.
11:26:06ddaltonby this one key press so its not a new screen...
11:26:14SlasheriJdGordon: at least voice callback isn't initialized without init
11:26:24Slasherithat might be enough to crash it
11:26:29Slasheribut i will fix that
11:26:29JdGordonah, yeah, maybe
11:26:48JdGordonI dont use voice, so thats probably how i missed it
11:27:17 Quit idnar (Nick collision from services.)
11:27:19 Join idnar_ [0] (i=mithrand@unaffiliated/idnar)
11:27:31JdGordonddalton: I have a patch for you to try
11:27:41ddaltonJdGordon: what is it for?
11:27:51JdGordonthe info screen
11:28:04ddaltonoh cool thanks where do I get it?
11:29:02JdGordonddalton:"> try it asap so i can commit it.... go into the info screen and move up and down to talk different items
11:29:24ddaltonok I will just give me some time to compile
11:29:29ddaltoncygwin takes for ever...
11:29:48jhMikeSddalton: I think if you enter the recording screen, the details of using the recording system are handled there. If you wish to enter/exit the recording screen with an action, sure it's possible like that.
11:30:39JdGordonddalton: actually.. dont worry, i have a voice file on my sansa.. it works goodly
11:30:55 Join ilgufo [0] (
11:30:56jhMikeSddalton: The recording screen obviously need special contextual behavior
11:31:09ddaltonJdGordon: can you commit I can't get it for some reason
11:31:19SlasheriJdGordon: yep, now working fine
11:31:19*ddalton Doesn't like ie
11:31:45ddaltonmaybe ff...
11:31:54JdGordonSlasheri: your fixing the simplelist stuff? can you add a call to the voice callback before the first draw?
11:32:07 Join lee-qid [0] (
11:32:10JdGordonotherwise the selected item isnt voiced on enter which is annoying
11:32:19jhMikeSbtw, I'd like a feature feeze on mpegplayer. It's going down the road of playback.c times 10.
11:32:37 Quit ompaul (Client Quit)
11:32:41SlasheriJdGordon: yep. hmm
11:33:17linuxstbjhMikeS: I thought there was just one feature being worked on - seeking?
11:33:18jhMikeSI'm going to rearrange it and police the structure. No stream should care about any other stream.
11:33:32SlasheriJdGordon: in fact i have no voice working either, but it seems just that the missing init was crashing it
11:34:01JdGordonbut you probably have voice menus enabled which would be enough to do it
11:34:07jhMikeSI can help that but it needs a stream manager in place that the UI uses and threads that have robust control. I'm doing that now.
11:34:19*JdGordon blames the upcoming delta on ddalton (unless its green :D )
11:34:23ddaltonJdGordon: is disk size and all of that in the list?
11:34:35JdGordoneverything that was voiced before is voiced now
11:35:00jhMikeSThe current implementation is very twisted
11:35:17ddaltonyes but stuff like disk size is part of the list is it?
11:35:25 Quit amiconn (" HydraIRC -> <- Would you like to know more?")
11:36:07ddaltonok well blind users scroling over that may find it annoying since it won't speak. Just commit and I will write a patch. I have a few here so just give me a bit to resync
11:37:04jhMikeSI've got the stuff basically in place.
11:37:10ddaltonamiconn: did you last commit touch thred.c?
11:37:21ddaltonhis gone
11:37:42ddaltonjhMikeS: were you talking to me?
11:38:17jhMikeSddalton: no, sort of just spilling what I'm working on with mpegplayer myself
11:38:22JdGordonohhh.... amiconn is in trouble :D lots of red
11:38:41ddaltonyes I saw that (can't compile)
11:38:59ddaltonwonder why I didn't see the red :-)
11:39:56ddaltonJdGordon: How the hell do I compile?
11:40:03jhMikeSI told amiconn to add #if NUM_CORES > 1 ! :p
11:40:19*jhMikeS will fix red
11:40:46 Join Juice^ [0] (n=juice@
11:43:39jhMikeSok, committed
11:46:22ddaltonJdGordon: will test and write up a patch tomorrow if I can work out how. Just about to attempt a linux installation
11:47:51 Join webguest33 [0] (i=dcef09c8@gateway/web/cgi-irc/
11:48:51 Quit webguest33 (Client Quit)
11:48:57 Join barrywardell [0] (n=barrywar@
11:50:09 Nick fxb__ is now known as fxb (
11:50:46 Quit ilgufo ("So Long, and Thanks For All the Fish -")
11:51:26 Nick bb_ is now known as bb (n=bb@unaffiliated/bb)
11:58:18 Join amigan [0] (i=dcp1990@unaffiliated/amigan)
11:58:23ddaltonJdGordon: just tested and needs to voice the other options otherwise will be to confusing. I will add them tomorrow
11:59:15 Quit kclaf (Read error: 113 (No route to host))
12:05:19ddaltonJdGordon: what is case 1 really?
12:05:24ddaltonin main_menu.c?
12:05:29ddaltonre the info screen patch
12:08:50daurnimatorhey all
12:08:55daurnimatorhow is the mrobe port going?
12:12:26 Quit FOAD ("I'll be back")
12:15:30 Quit Juice^ (Read error: 104 (Connection reset by peer))
12:16:44jhMikeSJdGordon fried his m:robe I think... :(
12:17:35 Quit atsea-22 (Remote closed the connection)
12:18:13*jhMikeS wonders how to hide a hardware decoder under libmpeg2 calls
12:19:00 Quit ddalton ("leaving")
12:20:15 Quit bertrik (Read error: 110 (Connection timed out))
12:24:29markunjhMikeS: which player has hardware decoding?
12:24:45jhMikeSmrobe should
12:25:16markunseparate chip or inside the DM320?
12:25:36jhMikeSa separate core inside the DM320 iiuc
12:27:02jhMikeSthat's what I gather from some briefs. I don't know if we know how to use it or if some info is available somewhere from some chip that is publicly documented.
12:27:14 Quit barrywardell ()
12:27:38 Join stripwax [0] (
12:28:35stripwaxello. just wondering if anyone has taken a look at dissassembling the apple diag mode to see if there's any power saving stuff going on in there (guessing that diag mode turns off IDE power except for Hard Drive Diagnostic test, etc.. )
12:29:08stripwaxI'm doing a noddy runtime test on diag mode (turn on fully charged ipod 5g, go into apple diag mode, .. leave it on until it dies)
12:29:56 Nick parafin|away is now known as parafin (
12:30:44JdGordonjhMikeS: na it just overheated in the weather and battery went flat... all good again
12:33:24jhMikeSJdGordon: must be hot down there
12:34:07stripwaxguessing linuxstb must be looking at something similar, given flashsplit.c on the wiki
12:34:27JdGordonjhMikeS: got to 37 i tihnk today
12:34:30JdGordonspring be here :D
12:34:42jhMikeSNice and cool here now :)
12:35:55*JdGordon back in a few hours
12:36:14 Join bertrik [0] (
12:38:55daurnimatorJdGordon: really?
12:39:16daurnimatorjhMikeS: use neuros's closed source decoders - they will work on the dm320 :)
12:40:10 Join lee-qid_ [0] (
12:40:33daurnimatorI'm serious - there is no compiler for the dsp
12:40:52daurnimatormust use binarys
12:41:08jhMikeSthey don't even know how to program it?
12:41:11 Join keanu|away [0] (n=chatzill@unaffiliated/keanu)
12:41:31daurnimatorwell, archopen got audio output going with a bit of asm
12:41:49 Join kclaf [0] (i=kclaf@
12:42:17jhMikeSwell, if info can be gotten, modify gas to do it.
12:43:10jhMikeSgcc assembler
12:44:08 Join Frazz [0] (
12:44:13 Join linuxstb_ [0] (
12:44:24 Quit linuxstb (Nick collision from services.)
12:44:26 Nick linuxstb_ is now known as linuxstb (
12:44:30stripwax(backlight diag with backlight set to 0% to be fair..)
12:45:01linuxstbstripwax: Not really. My interest in the flash is eventually putting Rockbox there...
12:45:15stripwaxah ok
12:45:36stripwaxcould be a good source for any missing info ..
12:47:40 Join atsea-22 [0] (i=atsea-@gateway/tor/x-12bb8152488095d8)
12:56:56linuxstbZagor: (for the logs) I don't know if anyone else has mentioned it, but the script doesn't work any more for me - the ARM multilibs patch is 404
12:57:36 Quit lee-qid (Read error: 110 (Connection timed out))
13:02:34 Quit lee-qid_ (Read error: 110 (Connection timed out))
13:02:36jhMikeSlunxstb: was an IRAM stack important for buffering on coldfire in mpegplayer?
13:03:14jhMikeS*linuxstb - sorry
13:03:17linuxstbYou mean the buffering that happens in the main thread?
13:03:34linuxstbI never tested it outside the main thread.
13:03:35 Join Temik [0] (n=Artem@
13:03:44rasherlinuxstb: perhaps you should just get them fom the CrossCompiler wiki page, put them in svn and change to get the files from viewvcs
13:04:03jhMikeSI thought about using UI on the main thread and running buffering on its own but it could be reversed easily.
13:04:43linuxstbUI on the main thread would seem cleaner.
13:05:53jhMikeSI've already got that setup and it just controls the stream with simple mpeg_stream_play/pause/stop/seek calls.
13:06:36 Join spiorf [0] (
13:06:50linuxstbI would be surprised if an IRAM stack makes much difference to buffering. But if it did, it should only need a tiny stack.
13:07:03linuxstbYou could even declare that stack on the main thread's stack...
13:07:32jhMikeScarefully though, UI can use alot itself _or_ a stack trade is possible.
13:08:20 Join FOAD [0] (
13:08:21linuxstbjhMikeS: BTW, if you want an mpegplayer feature-freeze, you should tell roolku.
13:09:05jhMikeSI think I will. It's important the this not get all mixed-up and structure stay clean. It's quite complex in general.
13:10:46linuxstbbertrik: Do you know much about the Linux USB stack? I'm having trouble talking to a device, but it works fine (with the same code using libusb) in Mac OS X...
13:11:07bertriklinuxstb: not about the internals
13:11:12jhMikeSbasically, he should work on his algorithm but not dedicate anything to a particular structure
13:11:21bertrikyou use libusb?
13:11:38linuxstbYes, but libusb isn't the issue. Let me explain...
13:11:58bertrikmaybe it is, there are differences between windows/linux/mac
13:12:40bertriklike whether you need to set the configuration explicitly or not
13:13:01linuxstbI'm talking to devices with Telechips CPUs in "usb boot mode". When I attach the device to my Linux PC, it appears in /proc/bus/usb/devices, but doesn't appear in "lsusb". It seems the device is timing out when Linux is asking it for the configuration.
13:13:15 Join amiconn [0] (n=jens@rockbox/developer/amiconn)
13:13:44bertrikouch, sounds like a bug in the USB firmware
13:13:58linuxstbYes, but it works fine in Mac OS X...
13:14:08***Saving seen data "./dancer.seen"
13:14:22linuxstbAnd in Windows with the official code that talks to such devices in USB mode.
13:14:43 Join n1s [0] (
13:14:44bertrikdoes it make a difference whether do you "lsusb" or "lsusb -v"? the latter also asks for the string descriptors IIRC
13:15:11linuxstbYes, neither work IIRC. "lsusb -t" works, but I think that's just parsing /proc/bus/usb/devices
13:15:55linuxstbIf I bypass libusb and try and open() the usb device node, then the open fails with (I think) E_TIMEOUT
13:16:09 Join ddalton [0] (
13:16:15bertrikdoes /var/log/messages or dmesg report anything suspicious?
13:16:52linuxstbI can see the device connecting. Then when I run lsusb, the device disconnects and reattaches.
13:17:38 Quit davina (Remote closed the connection)
13:19:21linuxstbThis has been tested with 3 different USB devices, each with slightly different Telechips CPUs. One of them works and shows up in a normal lsusb (iaudio 7), but the other two are exhibiting this behaviour.
13:20:26bertriklinuxstb: i don't think i can help you with this, sorry
13:20:46linuxstbOK, thanks for listening ;)
13:20:58bertrikit's either a bug in the firmware of the device or a problem really deep in the linux kernel
13:21:32linuxstbI guess I should find a linux usb mailing list and email them.
13:21:52bertrikyeah, I think there's a linux-usb-devel and a linux-usb-users list
13:22:23amiconnjhMikeS: Euh, sorry for the red, and thanks for fixing.....
13:22:37bertrikthey're on sourceforge: / users
13:22:57 Join Nico_P [0] (n=nicolas@rockbox/developer/NicoP)
13:23:04jhMikeSamiconn: np
13:24:01 Join Orc [0] (n=Orc@
13:24:20rasherSlasheri: what about enabling auto update of deleted files on non-flash targets as well? Do you want to add a warning in the ui first?
13:24:32 Join Rincewind [0] (
13:25:57 Quit safetydan ("Leaving")
13:26:01*amiconn would rather like to see some database bugs fixed :/
13:26:22 Join davina [0] (
13:26:47rasheramiconn: to be fair, enabling this is pretty much just changing a #define (except for the warning ui)
13:27:11amiconnI would not want auto update enabled on archos, btw
13:27:50amiconnBut I know of at least one 100% reproducable bug, and another one that was reported many times but might not be 100% reproducable
13:27:53rasherThen don't enable it?
13:28:32rasherI'm just talking about making Auto update work as you'd expect it (updating both added, modified and removed files) - even if it's rather slow.
13:28:48rasherCurrently it won't detect removed files unless dircache is enabled
13:28:55amiconnI know
13:29:07rasherWhich is very odd behavior
13:29:28 Quit Temik ("Leaving")
13:29:55Slasherirasher: i am going to add a warning
13:30:05Slasherirasher: and in general improve tagcache error reporting
13:30:23rasheramiconn: what do you mean, then?
13:30:32Slasheriamiconn: what kind of bugs?
13:30:47amiconnSlasheri: The 100% reproducable bug is that database will never finish its init and try over and over if there is no music file present
13:30:52Slashericurrently i know no bugs in the engine itself
13:30:53Orcmay I ask question? soory for my english
13:31:03Slasheriamiconn: ah, that one
13:31:07amiconnI already reported this several times
13:31:10 Quit stripwax ("Miranda IM! Smaller, Faster, Easier.")
13:31:18n1sOrc: shoot
13:31:30Slasheriamiconn: i will check that soon
13:31:43*amiconn should really use the tracker to report such things
13:31:50OrcHow match time to go, before D2 realese
13:31:54 Quit ddalton ("leaving")
13:32:11Orcmounth? 2 mounth or more?
13:32:16n1sOrc: no one is working on a port afaik so it is impossible to say
13:32:18rasherOrc: Impossible to tell, though probably a long time.
13:32:28 Quit Xerion (" ")
13:33:17n1sSlasheri: do you know about the endless include loops possible by including tagnavi.config in tagnavi_custom.config? like in FS #8000
13:33:18 Join agm3nt [0] (
13:33:18linuxstbOrc: If you've been reading my forum posts, that's just trying to get a tool working that will open up the possiblity of a D2 port. It still needs someone who owns a D2 to step up and do the hard porting work.
13:33:31 Join Domonoky [0] (
13:33:38Slasherin1s: yep
13:34:00n1sok, might be nice to add a check for that too :-)
13:34:19Slasherin1s: but i am not sure what should be done for that kind of thing
13:35:29n1sparseing of tagnavi.config is hard-coded, right?
13:35:31 Join moos [0] (
13:35:42n1sI mean that it is always parsed
13:36:11n1sso we could check if it is included anywhere and skip that include as it is never correct afaiu
13:37:02 Quit Orc ("Leaving")
13:37:09amiconnjhMikeS: Do you have objections against me introducing a generic "HAVE_YUV_DITHERING" macro in lcd-target.h ?
13:37:21rasherIt's either that, or keep track of all files parsed, and check for each include if it's already been parsed
13:37:58amiconnMeh, lcd-target.h doesn't exist...
13:38:21jhMikeSamiconn: no if you think some targets just won't ever implement it
13:38:36jhMikeSgray already does by default anyway
13:39:05jhMikeSdifferently of course
13:39:22 Join Temik [0] (n=Artem@
13:40:02 Quit FOAD ("I'll be back")
13:40:34 Quit linuxstb ("ChatZilla [Firefox]")
13:40:46 Join Nick_Brackley [0] (
13:40:53jhMikeSit comes to mind that perhaps people might want movies in black and white if the target is slowish? I'm sure chroma decoding could be skipped.
13:40:53 Nick Nick_Brackley is now known as darksaboteur (
13:41:19 Join jchord [0] (
13:41:41amiconnIt would at least make sense to skip chroma decoding on greyscale targets if libmpeg2 allows it
13:42:32jhMikeSyes of couse. I think it needs to be made to ignore chroma. There's no switch built in.
13:45:09 Join pixelma [0] (n=Marianne@rockbox/staff/pixelma)
13:48:04 Quit jchord ()
13:49:55 Quit grndslm ("Leaving")
14:12:06 Join Honkboy [0] (
14:19:00bertrikhas any reverse engineering been done on battery charging a sansa e200?
14:23:27 Part Honkboy ("Leaving.")
14:24:43jhMikeSbertrik: none that I'm aware of
14:25:01 Join Xerion [0] (
14:25:01 Quit kclaf (Read error: 113 (No route to host))
14:27:00bertrikwhat is the current charge "algorithm" then? just simple constant-current with a low current?
14:28:01 Quit cadu (Remote closed the connection)
14:28:24*bertrik is checking the source
14:29:13rasherIsn't it handled in hardware?
14:31:30jhMikeSbertrik: it's at least partly handled by the AS3514. some software control to at least switch it on and set voltages and current.
14:31:36bertrikI'm not sure. From the PP5022 product brief I see there is a fixed trickle charge mode, a programmable constant-current mode and a programmable constant-voltage mode
14:32:04bertrikOh it's done by the AS3514, not the PP5022?
14:32:30jhMikeSthe AS3514 handles tons of stuff on PP5024
14:33:40jhMikeSBesides having an audio codec, it's rather analagous to a PCF50606
14:36:10bertrikWould be nice if the hardware handled, so software can't accidentally kill the battery
14:37:59amiconnjhMikeS: I'm slightly confused about the scaling of r, g and b in your lcd_write_yuv420_lines_odither()
14:39:13amiconnUp to line 383 I am able to follow, but after that I am confused
14:39:20bertrikis the AS3514 part of the PP5024 chip, or is it separate?
14:39:43amiconnWhy is the delta scaled differently for B and R? Both have 5 relevant bits
14:40:20 Join kugel [0] (i=kugel@unaffiliated/kugel)
14:42:23jhMikeSbertrik: it mostly is
14:42:41jhMikeSbertrik: and is is part of the PP5024
14:43:15jhMikeSamiconn: to put thing in place so the result ends up at the correct position. did the comment get lost?
14:43:53jhMikeStherwise I'd need an additional instruction
14:44:04bertrikok thanks, i also see in the debug menu that the ADC channel for battery channel has already been determined
14:44:30amiconnAh, you mean "Red scaled at twice g & b"?
14:44:50amiconnFor c200 I don't need that
14:45:11J3TC-The sudoku patch didn't work
14:45:32amiconnI have to write byte wise
14:45:39jhMikeSamiconn: yes
14:45:55amiconnIs that easy to change? (the scaling I mean)
14:46:09amiconnI don't technically need to, but it might be less confusing
14:46:26amiconnOr should I leave it that way to keep similarity between targets?
14:46:54amiconnH10 and nano would make use of that trick....
14:49:18 Join stripwax [0] (
14:49:52jhMikeSI guess whatever results in the fastest code for any LCD would be the way to go. Different scaling may need different mults but not for R. If I actually mutiplied at that precision, it would have been one more addition.
14:51:03jhMikeSjust shift all R calcs by 1 less
14:51:26amiconnWell, instruction count won't change by this
14:52:11n1sare there any good getting started guides for m68k/coldfire assembly?, all I can find is for x86 :-/
14:53:19jhMikeSIt saved a shift during pixel packing. I only needed to mask.
14:53:33amiconnYes, saw that
14:53:34 Quit Rincewind (Read error: 104 (Connection reset by peer))
14:53:46amiconnBut for c200 I need to pack into 2 bytes
14:54:28bertrikoooops, battery voltage was 4.4V just now
14:54:35 Join lazka [0] (
14:55:37*jhMikeS is glad for bertrik that the battery is easily replaced :)
14:55:58stripwaxn1s - coldfire docs and rockbox source code make for a pretty good combination :)
14:56:15stripwaxespecially (i think?) some of the lcd code that exists as both coldfire asm and straight c
14:56:27J3TC-When I do a which command on m68k, it says command not found
14:56:46J3TC-m68k-elf-gcc and sh-elf-gcc
14:56:51n1sstripwax: that's what I have been looking at but most of the rockbox code doesn't have enough comments for an asm newbie...
14:57:22stripwaxoh.. :(
14:57:44n1sbut yeah, I'll dig and trial &error a bit more :)
14:58:04n1sJ3TC-: have you added the compiler dir to your PATH ?
14:58:26stripwaxn1s if it helps there's a coldfire emulator out there to test+learn with prior to crashing your audio player :)
14:58:45n1sstripwax: link or name?
14:58:53stripwaxI think it's linked from the h1xx wiki page but I'll see if i can find a link
14:59:02J3TC-export PATH=/usr/local/arm-elf/bin:$PATH
14:59:05J3TC-Yeah I typed this
14:59:45n1sJ3TC-: if you want to use m68k tools you need to add those too, you just added arm tools to the oath
14:59:48bertrikoh battery voltage is only that high during charge, the OF still thinks it needs more charging
14:59:49stripwaxn1s here
15:00:19n1sstripwax: thanks :_)
15:00:52stripwaxand as that page says, the emulator is 'incomplete' and certainly doesn't behave exactly like any iriver, nor does it emulator the h1xx cpu (emulates a prior model)
15:01:09kugeljhMikeS: around?
15:01:35stripwaxi think we stopped looking as soon as linusn weird up his jtag cable to his iriver..
15:01:45stripwax^weird^wired , heh
15:01:48jhMikeSkugel: I think so
15:02:14kugelsomeone said i should talk with you about FS #7999
15:02:47*jhMikeS looks at FS #7999 to find out what that is (haven't checked FS mails in a bit)
15:03:04 Quit agm3nt ()
15:03:46jhMikeSI think that may be fixed. A small error in thread.c was just fixed.
15:05:25 Join kclaf [0] (i=kclaf@
15:05:28*amiconn got funny looking colour bands when trying dithering on c200
15:05:30 Join PaulJam [0] (
15:05:37kugelok, I'm trying the latest current build
15:05:38bertrikkugel: it seems to be fixed for me with the latest SVN
15:05:51bluebrotherkugel: there was a fix about this issue today, check the svn logs
15:05:52*jhMikeS got lots of funny looking things writing the dithering :)
15:06:11bertrikconsider it a feature, not a bug :)
15:07:22*jhMikeS hates those little dumb mistakes (forgetting to reenable interrupts) :p
15:07:49 Quit kubiix ("Miranda IM! Smaller, Faster, Easier.")
15:08:18jhMikeSthough I'd say for several thousand code lines it's been relatively trouble free
15:12:01 Quit Temik ("Leaving")
15:12:15kugelseems to be fixed
15:12:26jhMikeSkugel: so I can close #7999?
15:12:48kugeli think so
15:12:53jhMikeSok, cool
15:14:10***Saving seen data "./dancer.seen"
15:18:42 Join ompaul [0] (n=ompaul@gnewsense/friend/ompaul)
15:19:29 Join midgey [0] (
15:19:29 Quit mf0102 ("Verlassend")
15:23:32*JdGordon returns
15:24:30JdGordonSlasheri: HZ/100 is a bit overkill dont you think?
15:26:07stripwaxwhich has a better z80 core implementation, pacbox or zxbox or rockboy or .. ?
15:26:39jhMikeSumm...sure HZ/10 wasn't meant?
15:27:56stripwaxthinking pacbox probably not the fastest, and if so it would be nice to port one of the others over and see if audio playback is possible during pacman..
15:28:21jhMikeSit is on gigabeat but that's because no IRAM is used there
15:28:35 Quit darksaboteur ("ChatZilla [Firefox]")
15:29:33stripwaxwhich is what?
15:29:44stripwaxpacbox *is* fastest on gigabeat?
15:29:58stripwaxor audio playback isn't disabled on gigabeat?
15:31:14stripwaxI'm wondering if we could reduce iram usage on ipod 5g for example by unifying those thousand-odd opcode functions into several core functions that fit inside instruction cache
15:32:07*amiconn found the bug in c200 yuv dithering :)
15:32:15n1sstripwax: to be able to have playback while in the plugin you can't use _any_ iram in the plugin
15:32:16jhMikeSstripwax: plugin IRAM and codec IRAM are shared. It's special highspeed SRAM but it's quantity is limited.
15:32:18SlasheriJdGordon: not for synchronous db scan, actually it's a bit slow still :)
15:32:32stripwaxn1s/jhMikeS - yep, I know that.
15:32:35JdGordonI dont tihnk it needs to be that fast....
15:32:46stripwax(by reduce I suppose I meant eliminate, sorry wasn't clear)
15:32:46 Quit XavierGr (Remote closed the connection)
15:32:49JdGordonthat 1 debug screen doesnt _have_ to be ealtime
15:33:00 Join XavierGr [0] (
15:33:01SlasheriJdGordon: well, with HZ/100 it can scan 50 files/s at max
15:33:02 Join lee-qid [0] (
15:33:11jhMikeSyeah, so PLUGIN_IRAM_INIT on gigabeat is just #defines as empty there.
15:33:16Slasheriat least on flash devices that would be slow
15:33:20JdGordonit does the scan in that callback?
15:33:33SlasheriJdGordon: no, but it's synchronous
15:33:37stripwaxdo the other z80 emulator cores use iram?
15:33:43Slasheritagcache waits for the callback
15:34:18JdGordonthere isnt a "better" way to do it?
15:35:17SlasheriJdGordon: debug menu must be realtime if we want to find out on which file tagcache crashes
15:35:36Slasheriof course that value can be turned down but it makes the scan slower
15:35:37JdGordonok, my next commit will fix you up a bit then
15:35:44jhMikeSstripwax: I havent't looked at them but I'd be surprised if they didn't at least for coldfire.
15:35:44bluebrotherSlasheri: > Hi, all.
15:35:44bluebrother> This is rather a large favour to ask for, but I hope someone would
15:35:44bluebrother> be good enough to manage it. Could anyone with the Toshiba Gigabeet F40
15:35:54bluebrotheroops, sorry.
15:36:35JdGordonSlasheri: how about TIMEOUT_NOBLOCK for the db debug screen then?
15:36:36jhMikeSactually, IRAM doesn't help much with a data cache. heck I've even seen it run more slowly.
15:36:43bluebrotherSlasheri: how about splitting the file display up into path and filename? Currently it's kinda impossible to read the filename on small displays
15:37:39Slasheribluebrother: hmm, true
15:37:46SlasheriJdGordon: no idea, please try out :)
15:38:22amiconnjhMikeS: There's a strong indication that the lcd bridge is the bottleneck on c200:
15:38:45amiconnlcd_yuv_blit() achieves the *exact same* fps as lcd_update() for fullscreen
15:38:45stripwaxjhMikeS - yeah, thinking on ipod at least the data+instruction cache would help out at least as much here. ok, will see what the other implementations have to offer..
15:38:55stripwaxamiconn - wow!
15:39:14amiconn...and 1/4 updates are only slightly slower - by about 1%
15:41:56 Quit davina (Remote closed the connection)
15:43:17 Join FOAD [0] (
15:43:48stripwaxanyone had any thoughts about the bitmap sliding puzzle (FS #7218) since, ehm, May?
15:43:59 Join davina [0] (
15:45:25JdGordonSlasheri: it seems to be updatng alot faster if the debug screen isnt open?
15:47:43jhMikeShow 'bout TIMEOUT_NOBLOCK + yield() ?
15:48:29JdGordonif a screen sets timout_noblock then its up to it to yield..
15:48:59jhMikeShence the
15:49:19 Part pixelma
15:49:32jhMikeSnow, I wonder if I should force yields when queues are used with timeout = 0
15:50:45JdGordonwhat happened to going premetive so we didnt have to wrry about yielding?
15:51:26jhMikeSyou think I was going for that right now? :) that yield would sort of be preemption.
15:52:11stripwaxor even FS #7497 for that matter. wasn't this a requirement for getting bootloader versions working in rbutil?
15:52:16*stripwax shrugs
15:54:11jhMikeScommunicating with any UI-creating thread through it's own message queue would help in a preemptive environment. would even be good in a touchscreen environment.
15:55:00jhMikeSthat leads to an easy way to have real windows with focus (even better with a touchscreen) :)
15:55:32stripwaxDomonoky - can you recall? didn't you need a specific magic string to be present in the bootloader file somewhere on-disk in order for rbutilqt to work out what version bootloader was installed?
15:56:11bluebrotherIIRC the bootloader version is currently only determined by the log file
15:59:19stripwaxand if there's no file? i recall the idea was to look for the bootloader file on-disk (for those devices for which this is possible) and see if the magic string was present and use that to determine the bl version number
15:59:42stripwaxbut it seems patch isn't committed so I guess that's not possible
16:02:26bluebrotheron Ipod (and Sansa?) it would be possible to read the BL from disc and check / md5sum (my favourite as it would also verify the integrity) it
16:05:11SlasheriJdGordon: of course, then tagcache doesn't need to wait for screen updates
16:05:15stripwaxhm, so rbutil would need to download a new list of md5sums every time there's a new bl build. I think we discussed that last time round and decided a magic string would be better! :)
16:05:44stripwaxI'm easy either way, would be nice if someone could commit some of these patches that other people have asked for :)
16:05:59 Quit keanu|away ("ChatZilla [Firefox]")
16:06:55bluebrotherrbutil currently downloads a list of md5sums anyway ;-)
16:07:08bluebrotherbut I agree that it would be better to have this network independent.
16:07:26bluebrothermaybe extend the magic to add the md5sum of the bl bin?
16:07:33 Quit ompaul (Client Quit)
16:07:39stripwaxif i do that is it any more likely to get committed at any point? :-)
16:09:01 Join guywithproblem [0] (i=52aea24a@gateway/web/cgi-irc/
16:09:03stripwaxbtw what's the advantage of putting the md5sum into the file automatically? is it just to protect against the (rare?) case of someone having a bootloader bin that looks like a rockbox bootloader and has the rbblver magic but is somehow corrupted?
16:09:20stripwaxthinking that if someone builds their own customer bl then the md5sum would still 'match' so not sure what it guards against
16:09:46stripwax(^corrupted^modified by some third party thing we don't know about yet)
16:09:59bluebrotherwell, it's nothing that is needed in any way of course.
16:10:11guywithproblemhey everybody im testing rbutil on mah vista os everything works fine exept the downloading of rockbox i get http error C:\Users\kevin\Desktop\rbutilqt-v1.0.2
16:10:13bluebrotherbut I like the idea of having it in the same file.
16:10:36bluebrotherguywithproblem: have you checked the proxy settings?
16:10:58guywithproblembluebrother : yup i putted no proxy
16:11:34stripwaxbluebrother - still not sure I understand. put the md5sum of the bin file, into the bin file, .. and then what? if you then md5sum the file you'll get something different. or if you md5sum everything except the md5sum-in-the-file you'll get the same. and then what ?
16:12:03bluebrotherand you're sure your network is working? Can you access from your browser?
16:12:20bluebrotherstripwax: well, my idea was to have two different formats −− like bl.bin and bl.ipod
16:12:33guywithproblembluebrother Yes i just downloaded the util XD
16:12:36n1sstripwax: quick cf asm question, do I need to save all registers? most functions don't save d0 d1 a0 a1 but all the others they use
16:12:40bluebrotherand add some magic including the md5sum to the latter.
16:12:50bluebrotherbut maybe that idea wasn't really good at all ;-)
16:13:12bluebrotherguywithproblem: can you try it on a non-vista machine? Vista is currently completely untested
16:13:14stripwaxn1s - it's a matter of convention, usually d0,d1,a0,a1 are free to use a scratch by a function as they are not preserved
16:13:19guywithproblemah bingo ^^ its working fine on vista now :) there was a problem in the firewall setting :)
16:13:23stripwaxyou certainly need to store all others that you use
16:13:29guywithproblembluebrother IM testing on vista now :p
16:13:53bluebrotherit's known that bootloader installation fails on vista in case you get problems ;-)
16:14:08stripwaxn1s - let me find you a link
16:14:34guywithproblembluebrother nah i dont care its mah old ipod so cant complain if its broken :)
16:14:42stripwaxn1s this might help
16:15:20stripwaxgcc convention: " Register Usage
16:15:20stripwaxRegisters D0, D1, A0, and A1 are scratch registers and are not saved
16:15:20stripwaxand restored when calling other functions. All other registers that
16:15:20DBUGEnqueued KICK stripwax
16:15:20stripwaxare used by a function must be saved on the stack before being
16:15:20stripwaxmodified, and restored from the stack before the function returns."
16:15:34bluebrotherrbutil won't detect the ipod (or install the bootloader) on vista ... but ...
16:15:37CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
16:15:37*bluebrother got an idea
16:15:54guywithproblembluebrother tell me ?!
16:16:05stripwaxbluebrother/guywithproblem - if you right-click the Properties of rbutil and select to "Run as Administrator" does it work?
16:16:34bluebrotherstripwax: bootloader installation / autodetection for ipods / sansa doesn't work on vista due to UAC issues
16:16:35guywithproblemOw yeah stripwax/bluebrother i Disabled That Thing where do always need to click continue or yes or something :p
16:16:40n1sstripwax: thanks! that clears up a lot of questions :-)
16:16:48stripwaxn1s - cool!
16:16:58stripwaxactually there's a link to a pdf there, don't know it but might be a nice doc
16:18:17stripwaxbluebrother - even if you specify the 'run as administrator' preferential compatibility setting for the app?
16:18:42bertrikis there a smarter way of writing the rockbox package to the device than unzipping the archive to it?
16:19:04guywithproblemill keep you guys updates as Util is working :) currently downloading a bunch of Themes
16:19:05stripwaxbertrik- 'smarter' how - like only replace the files if they're newer/different?
16:19:16bertrikstripwax: exactly
16:19:44jhMikeSn1s: I could've told you all that :)
16:19:57stripwaxbertrik - hm, nothing I'm aware of. from experience though rbutil seems much faster at physically unzipping onto the device than the built-in Windows unzip, if you're using that.
16:20:12bluebrotherguywithproblem: you could try installing the bootloader −− put the file in the same folder as rbutilqt.exe and try if autodetection of the ipod works
16:20:20stripwaxbertrik - but potentially every file could change in some way between versions, and rockbox builds don't track the dependencies really
16:20:20n1sjhMikeS: I'm sure I can come up with some other question to nag you with :-)
16:20:33jhMikeSok...challenge me :)
16:20:47guywithproblembluebrother Its okey i forgot i already installed it once for the ipod games hack :p
16:20:59bluebrothernot sure if this is suffcient, haven't really looked into this manifest stuff at all
16:21:11 Join ompaul [0] (n=ompaul@gnewsense/friend/ompaul)
16:21:26bertrikstripwax: ok
16:21:27bluebrotherwell, it would tell me if the manifest thingy is that easy or not ;-)
16:21:27guywithproblembluebrother :) ill try it on the other vista with my other ipod XD
16:21:49bluebrotherit also doesn't matter if you have installed the bootloader before or not
16:22:38guywithproblembluebrother Good to know
16:22:38 Join Lear [0] (i=chatzill@rockbox/developer/lear)
16:23:53J3TC-If a hunk failed during patching then the patch didn't work right?
16:24:17Nico_PJ3TC-: it's partially applied is but the whole patch itself is useless right?
16:24:38 Join Buschel [0] (
16:24:59stripwaxJ3TC- most probably
16:25:16stripwaxunless it's easy for you to manually apply that failed hunk (look for the .rej file)
16:26:27guywithproblemWtf that manifest made the players list empty XD
16:27:47J3TC-Ok, what if you applied a crapload of patches and but you want to revert back to the original svn
16:27:49bluebrotherthe list is now _empty_? Wow, that's interesting.
16:27:52J3TC-Do you just do svn up?
16:28:23krazykitJ3TC-, you could use patch -R
16:28:34J3TC-Yeah but I get hunks failed there too
16:28:35bluebrotherand if that fails svn revert -R .
16:29:22J3TC-It says not enough arguments
16:29:29J3TC-Ah nvm
16:29:33 Join Gnu47 [0] (
16:29:34J3TC-Forgot the dot there lol
16:29:45guywithproblemokey :p that ipod is FFd up now XD
16:29:56J3TC-I'm just applying patches and see what works and what doesn't and get a list going
16:30:09 Quit JdGordon (Remote closed the connection)
16:30:13J3TC-So when I get a good list, I could just apply straight through
16:30:48*jhMikeS hates circular disk buffers. Must've had a traumatic childhood experience with one.
16:31:57*guywithproblem hates that he needs to work in a server room
16:32:04n1smy second m68k ams function that actually does something works :-) and the first one where I actually understand what i did...
16:32:24 Join Arathis [0] (
16:32:25jhMikeSn1s: for midi player?
16:33:01n1sjhMikeS: no, this one was for tremor, might very well be slower than c too but I think I will try to make something for midi :-)
16:33:17jhMikeSemac all the way on that
16:35:38guywithproblemrockbox is awsomm :p
16:36:57 Quit PaulJam (".")
16:37:35guywithproblemits working now ^^
16:44:17 Join webguest63 [0] (i=48d1e439@gateway/web/cgi-irc/
16:45:00 Quit webguest63 (Client Quit)
16:45:46 Nick guywithproblem is now known as happyguy (i=52aea24a@gateway/web/cgi-irc/
16:47:03*jhMikeS doesn't want to add another thread slot on SWCODEC ... argh ... oh well I need four free with voice on
16:47:48*jhMikeS takes that back :) it's only 3 - jolly good
16:47:51Nico_PJ3TC-: you can use the −−dry-run switch of patch to see if a patch applies cleanly without actually applying it
16:48:21Nico_PJ3TC-: also reverting a patch that failed to apply will fail at the same place so generally it's OK
16:48:57Nico_PjhMikeS: buffering needs another thread... is that a problem ?
16:49:04jhMikeSit does?
16:49:20Nico_Pwell there's a new thread involved, so yes
16:49:35jhMikeSmpegplayer will only need 3 additional ones to the main so 3 must be free to run it
16:49:45stripwaxn1s - i just successfully built that coldfire emu on cygwin, seems to still work (doesn't boot rockbox of course)
16:50:00stripwaxbut might be useful for tweaking/etc
16:50:07jhMikeSjust add another, no biggie
16:50:23n1sstripwax: nice, I'm not sure how to use it though...
16:50:26Nico_PjhMikeS: what do I need to change? a #define somewhere?
16:52:06jhMikeSheaps of IRAM were freed anyway in this kernel work and the slots aren't that big
16:52:33 Join mirak [0] (
16:52:58*Nico_P should think about IRAM usage...
16:53:08Nico_PI think I removed some vars that were stored in it
16:53:37n1sNico_P: have you found that mad skipping lockup bug?
16:54:41Nico_Pn1s: I've made changes that should make the lot a bit more solid, but haven't found the bug precisely... it not happening on my gigabeat doesn't help
16:55:03Nico_Pn1s: btw, would you have time to test the updated version ?
16:55:08n1sNico_P: I can test again a bit later today
16:55:10jhMikeSNico_P: frankly I don't think any target besides PP needs the thread/core strutures in IRAM. It's not accessed very much...even if it were, not much data is accessed.
16:56:39Nico_Phmm actually the only things I removed from iram are buf_ridx and buf_widx. long as codec on COP doesn't need to touch those
16:59:44 Quit midgey (Read error: 104 (Connection reset by peer))
17:00:13Nico_Pno, they are managed by the buffering thread now
17:01:17jhMikeSdoes codec read any of that? The buffer might need to use UNCACHED_ADDR because of handle contents. I don't want to cache flush/invalidate all the time there.
17:02:32*jhMikeS supposes that can probably be worried about later
17:02:33Nico_PI'm not sure I understand you fully, but the codec reads the data directly from the buffer by requesting a pointer to it with bufgetdata
17:03:18jhMikeSand the handles are on the buffer, right? not just the audio data.
17:03:24 Join Shaid [0] (
17:03:25Nico_Pyes, they are
17:03:52Nico_Pthe struct is right before the actual data in memory
17:04:30jhMikeS_could_ they be aligned and padded on some builds?
17:04:52Nico_Pthey are aligned on all builds currently
17:05:23Nico_Ppadded I have no idea, but everything uses sizeof so I think there should be no problem
17:05:24jhMikeSTo avoid cache line interference on PP between cores, things must be 16-bytes aligned
17:05:47jhMikeSexperiments seem to indicate - hope it's no larger
17:06:18Nico_PIIRC everything is 32 bytes aligned because of the gigabeat ATM
17:06:43 Quit happyguy ("CGI:IRC (EOF)")
17:07:22jhMikeSok, then we're might actually be that on PP as well. then some buffer data could be read and written at the uncached alias on PP.
17:09:16Nico_Pwhat does that mean ? (sorry I'm really not a low level guy)
17:10:14 Quit FOAD ("I'll be back")
17:11:21jhMikeSit means that reads and writes aren't held in the processor cache but instead are read and written directly in ram when the uncached alias is used. PP is setup to show both cached and uncached views of RAM depending upon the address offset used.
17:12:45 Quit Shaid ("I will see you, in the end. And I will laugh at your pain...")
17:12:47jhMikeSeach core has it's own cache and therefore their views of RAM will not always agree...hence the need for IRAM use in the kernel which isn't cached.
17:12:48Nico_Pwhat's better ?
17:13:20Nico_Pah ok I see
17:14:14***Saving seen data "./dancer.seen"
17:15:06Nico_PjhMikeS: is there a way to peek at a message in the queue without consuming it ?
17:15:53 Join FOAD [0] (
17:15:56jhMikeSNico_P: want one?
17:16:20Nico_PjhMikeS: it's not critical but it would be useful
17:16:21jhMikeSwhat do you want to know exactly? what the next message is?
17:16:50Nico_Pyes, I want to check it and maybe remove it if it's useless
17:17:02Nico_Pif it's not, leave it there to be treated
17:17:20jhMikeSthat would be really simple to add to kernel.c
17:19:10Nico_Phehe... I believe you but I'd feel a little incomfortable tampering with kernel.c... I can try though. queue_peek() ?
17:19:16 Join Morey [0] (
17:20:59jhMikeSI already implemented it...want a pastebin?
17:21:17Nico_Pah, yes please :)
17:22:07 Join andresmh [0] (
17:22:48MoreyjhMikeS: A few people have expressed the want to have a simple resume yes/no start menu for mpegplayer instead of the current one. If you get a chance check out #7997 on flyspray. This patch makes the start menu optional. Small patch should be quick to look at.
17:22:56andresmhI have an ipod 4th gen. and I want to install rockbox. But from this page I get that there is no rockbox for 4th gen *color* only grayscale, is that correct/
17:23:22Nico_PjhMikeS: do you plan on committing it or should it be part of my big commit?
17:23:40jhMikeSMorey: well, I am doing a reorganization of it that should make doing anything at all much easier
17:24:21jhMikeSNico_P: I hadn't planned on it for the moment. Will it be a huge deal to resync git if I do later?
17:24:45Nico_PjhMikeS: shouldn't be
17:24:48bluebrotherandresmh: the color is 4G
17:25:03bluebrother(video id 5G / 5.5G)
17:25:26andresmhmine is 20GB
17:25:51andresmhso what rockbox should install?
17:25:53bluebrotherbesides, rockbox utility / ipodpatcher will also display the Ipod it detected.
17:25:58andresmhthis ?
17:26:01Lloreanandresmh: The "iPod color/Photo" version is for all iPods sold as 'iPod Color' or 'iPod Photo'
17:26:31bluebrother :)
17:27:26 Quit Morey ("Ninja IRC v1.5.8.1(#1) exiting after 8m23s of use")
17:29:54 Join Limerr [0] (
17:30:12jhMikeSNico_P: though thinking about this, if you leave a message there, the queue will remain signaled until all messages are consumed. how to prevent pointless rechecking?
17:31:48jhMikeSof course "useless" messages could be pulled and placed in a lower priority queue on the side
17:33:46andresmhok, so I rbutilqt.exe didn't autodetect my ipod. Should I go ahead and install the bootloader or the audiplayer first?
17:34:28 Quit Limerr ()
17:34:52bluebrotherandresmh: do you have administrator permissions?
17:34:59Nico_PjhMikeS: what do you call a "useless" message ? one that isn't removed after queue_peek is called ?
17:35:51andresmhbluebrother: I am using winxp as admin
17:36:37bluebrotherhmm, sounds like the detection issue that has been reported on windows −− but I have no idea what's causing this.
17:36:54bluebrotheror have you installed any other software (like IPL) to the Ipod before?
17:37:09jhMikeSthe queue_peek I just doesn't remove anything. If you want it out, call queue_wait. It won't block you then.
17:37:19bluebrotherlooks like you need to use the command line ipodpatcher for bootloader installation
17:37:34andresmhoh, i think i messed up, i click on "install rockbox on your audio player" before installing the bootloader
17:37:45 Quit XavierGr (Read error: 104 (Connection reset by peer))
17:37:46Nico_PjhMikeS: ok. what's queue_remove_from_head for btw ?
17:37:47bluebrotherthat's not a problem.
17:38:10bluebrotherbootloader and build can get installed separately (well, as long as you selected the correct player ;-)
17:38:22jhMikeSNico_P: queue_remove_from_head removes all message with the specified id until it finds one with a different id
17:38:27andresmhi hope i did, i slected the apple/color ipod
17:38:47andresmhso how do I know if everything worked? should i disconnect the ipod from the pc? and turn it off/on?
17:39:13 Join ToHellWithGA [0] (
17:39:35webmindga ?
17:39:46bluebrotheryou need to install the bootloader first.
17:40:46andresmhbluebrother: done, i think it worked.
17:40:48jhMikeSNico_P: Of course if any of the message removed was sent, the thread waiting for a reply gets a default reply of 0
17:40:58andresmhnow i can disconnect?
17:41:07bluebrotherthen you need to reset the Ipod by pressing and holding Menu + Select
17:41:14bluebrotheryes, do a safe disconnect first.
17:43:25jhMikeSNico_P: It
17:43:58andresmhok, this is the first time i use an ipod, when i press menu+select it just shows up the ipod menu that was last used
17:44:04jhMikeS's also imperative that any queue with message sending enabled never allow multiple-thread access to the head. it should be considered owned.
17:44:10andresmhdoesn't seem to "boot"
17:44:22bluebrotheryou need to press and hold the buttons down until the display goes off
17:44:24 Quit mirak (Read error: 113 (No route to host))
17:44:54Nico_PjhMikeS: only the buffering thread is aware of its queue
17:45:06bluebrotherif it boots to AppleOS something went wrong with bootloader installation.
17:46:02andresmhbluebrother: ok, i got it to reboot, but it did boot to appleos :(
17:46:19bluebrotherdid you install the bootloader using ipodpatcher?
17:46:59andresmhi used the install bootloader option in Rockbox utility (rbutilqt.exe)
17:47:13bluebrotherand it showed success?
17:47:23andresmhdidn't say success nor fail
17:47:29andresmhso i clicked install again
17:47:37andresmhand it said somethinbg like "already installed"
17:47:54bluebrotherhmm. You said that rbutil didn't detect the Ipod. If that happens it can't install the bootloader.
17:48:06 Quit arkestra (" HydraIRC -> <- The alternative IRC client")
17:48:13andresmhbluebrother: but i selected the drive and then the ipod model manually
17:48:16andresmhnot good?
17:48:34bluebrotherno. It's ok for installing Rockbox but not for installing the bootloader
17:48:45Nico_PjhMikeS: what would queue_peek be like for the sim ? same thing without the IRQ and core_lock business ?
17:48:53bluebrothertry installing the bootloader using ipodpatcher as the manual installation instructions in the manual
17:49:09andresmhoh ok, i'll try that
17:49:11bluebrother(mostly run ipodpatcher and select "i"nstall)
17:49:46bluebrotherthere seem to be issues with bootloader installation on windows, but not in all cases. No idea what's causing it as rbutil uses the same code as ipodpatcher
17:50:04bluebrotherhopefully we'll figure out eventually.
17:50:30jhMikeSNico_P: yeah, exactly that
17:51:02SoapAve: You asked a question (around 36 hours ago) about replicating Doby NR's preemphasis through the Rockbox EQ for when you are connecting your player to a stereo system through the use of a cassette adapter.
17:51:10 Quit amiconn (Nick collision from services.)
17:51:17 Join amiconn [0] (n=jens@rockbox/developer/amiconn)
17:51:19andresmhbluebrother: ok, i downloaded ipodpatcher.exe and i reconnected the ipod to the pc. the screen of the ipod says "do not disconnect". i am going to execute the ipodpatcher.
17:51:22 Join FOAD_ [0] (
17:51:29 Quit FOAD (Read error: 104 (Connection reset by peer))
17:51:32 Nick FOAD_ is now known as FOAD (
17:51:44AveSoap: hi, yes. you have something on the subject?
17:52:16Aveafter examining the dolby working theory it seems that the emphasis on different bands is affected by amplitude of the signal also
17:52:17jhMikeSNico_P: It occurs to me it could return data about whether a thread is waiting on a reply as well. Well, I guess much _can_ be done if needed.
17:52:35andresmh[INFO] Wrote 5433344 bytes to firmware partition
17:52:36andresmh[INFO] Bootloader installed successfully.
17:52:38Avenot so simple as to just pre-emp high frequencies all over the place..
17:52:42SoapAve: It is my (incomplete?) understanding of the reasons behind Doby NR that it is to correct for a lacking of the media (here magnetic tape) not of the hardware (the reading heads). ASSuming this is correct you do not need to use Doby NR - as a cassette adapter should not suffer from the "hiss" Doby NR is ment to account for.
17:52:57andresmhbluebrother: reboot now?
17:53:03bluebrothernow reset the Ipod. Rockbox should load.
17:53:24AveSoap: ah ic ok that would make sense
17:53:50andresmhawesome, it worked
17:54:15andresmhso now i want to install a nicer theme. should i use rbutilqt?
17:54:37AveI suppose the ideal placement of the "heads" would be more important and somewhat difficult to achieve, also I dont think that the cassette adapter frequency response is particularly good
17:55:47andresmhhow do you boot into appleos?
17:55:47krazykitandresmh, it's up to you, but that's probably one of the simpler ways.
17:55:56SoapIt should, in theory, be far superior to an FM modulator. The S/N ratio isn't the best with tape, but the frequency response is excellent. Also, some of the nicer cassette-adapters I have seen in stores have an adjustable head.
17:56:01bluebrotherandresmh: flip the hold switch directly after poweron
17:56:06 Join Limerr [0] (
17:56:12bluebrotherand I'd suggest using rbutil for installing themes ;-)
17:56:21Nico_Pwow this makes playback super slow
17:56:24andresmhthanks guys, you are so helpful! :)
17:56:39bluebrotherI'd also suggest checking the manual −− it has a lot of usefull stuff in it :)
17:56:48AveSoap: heh ok I need to ddo some shopping then
17:57:00LimerrMy ipod photo is now officially retired, ipod video here i come :D
17:57:21 Join criznach [0] (
17:57:26LimerrWhich windows util is the best for installing rockbox onto the ipod video?
17:57:31AveSoap: thanks for your time, now I havent lost all my faith with the adapter approach..
17:57:40 Quit FOAD (Remote closed the connection)
17:57:55jhMikeSMorey: I thought about implementing a scaling YUV thumbnailer...perhaps with some interpolation to scale things down nicely for scene selection if desired. It can use lcd_framebuffer on color targets and could convert to RGB in place there. Speed isn't such a huge deal with that.
17:58:27SoapAve: It really is second best to Line-in, something to take into account, though, is you may need to drive it at -10dB or so. Mine distorts when driven @ 0dB.
17:58:55AveI've ran into that
17:59:05 Quit DataGhost (Read error: 110 (Connection timed out))
17:59:13Aveactually, I get massive distortion unless rb volume says "-30" or so
17:59:18Avedepends on the music
17:59:33Aveall eq settings at zero
18:00:08 Join roolku [0] (
18:00:22Avewhich is why I get into the noise problem, then I need to crank up the volume on the other end
18:03:02amiconnSlasheri: *Please* revert that insane get_action() polling in list.c. Doing that every tick is major overkill and will hurt all targets except those with very fast lcd updated
18:04:11SoapI'm sorrry for leading this sooooo offtopic, Ave. This really should be finished in #rockbox-community.
18:04:36LimerrWhile i am trying to install rockbox on my ipod 5.5 with rbutil i get http error 7602289
18:04:46 Join advcomp2019_ [0] (n=advcomp2@
18:04:49Aveoh yeah
18:04:55LimerrDoes anyone have a clue what might cause that?
18:08:01 Join Konam [0] (
18:10:24 Join mo-seph [0] (
18:10:27Konamam I the only one getting some "DJ disc ripping" sounds while playing mp3s on an ipod nano 1GB first gen
18:10:53Konamsorry for the "DJ disc ripping" line, I don't know how to say that in english
18:11:32Konamthey're annoying, it start playing ok by after a while it starts with those sounds
18:11:38 Join ina84 [0] (n=opengurl@
18:12:58 Quit stripwax ("Miranda IM! Smaller, Faster, Easier.")
18:14:18*ina84 slaps idnar_ around a bit with a large trout
18:14:26ina84ops sori
18:14:42SoapKohlrabi: That sounds similar to how my Nano starts to glitch before crashing. It is known that /some/ of the Nanos out there are suffering from an unquashed bug. When was the last time you updated your Rockbox build before noticing this problem?
18:14:55Soapoops, that was ment for Konam
18:15:17ina84hi guys, just wanna ask noob q: how to display lyrics in rockbox?
18:15:58KonamSoap I updated to today's build
18:16:03Soapina84: there is not an ability to display lyrics with Stock Rockbox. I do not know if there is a patch in the tracker to do such a thing - you should search flyspray and see if anyone is working on such a thing.
18:16:04ina84i use ipod 5.5g, and the recent rockbox version
18:16:06Konamand I'm still getting it
18:16:19SoapKonam: I grok that - when was the last time you updated before today?
18:16:23KonamSoap it does crash too
18:16:37 Join FOAD [0] (
18:16:43ina84ok tq
18:17:20 Quit Xerion (Read error: 104 (Connection reset by peer))
18:17:23 Quit davina ("xchat on Ubuntu 7.04")
18:17:55LimerrWhile i am trying to install rockbox on my ipod 5.5 with rbutil i get http error 7602289, anyone know how i can fix that or can recomend another installer?
18:18:28SoapLimeer - you can always use the method I believe is still outlined in the manual.
18:18:44KonamSoap oh, the build I use to have was this: Version: r13711-070625
18:19:17LimerrCan i still get a boatloader compatible with both iPL and Rockbox?
18:19:29SoapKonam: yep that is before the bug crept in. There is a forum thread of people discussing the bug - problem is none of the developers have a Nano which suffers from the problem.
18:19:29KonamSoap I updated a few days and those glitch started. Now I found my old build, I will try it and see if this happens there too
18:19:34DomonokyLimerr: sounds like network problems.. which os and player do you use ? (and did you use completeInstall in rbutil? )
18:19:55krazykitLimerr, the rockbox bootloader will boot both, and how should be outlined in the manual (or maybe the wiki)
18:20:02Limerrcomplete install, windows xp, no firewall what so ever :S
18:20:31SoapLimerr: The Rockbox bootloader is perfectly capable of loading IPL. Use of the IPL bootloader, while supposedly compatible with Rockbox, is not supported due to it not being Rockbox code, and issues with wether it initializes the hardware properly.
18:21:06Limerrk, i will try installing it by the wiki-manual then :)
18:21:10KonamSoap damn, but, they simply have to check what changes they've made and try to revert them...
18:22:08KonamSoap do you think it is a codec thing?
18:22:32 Quit advcomp2019 (Read error: 110 (Connection timed out))
18:22:42DomonokyLimerr: try it with rbutil, but not the complete Install , use the Install tab, and install the things seperatly
18:22:47 Join MethoS- [0] (
18:23:31Domonokyif fixed a severe bug in complete installation of rbutil, but this isnt in the newest rbutil release... :-)
18:23:36SoapKonam: it is not as simple as that. The change which appears to have _exposed_ the problem is one which will not be reverted. The change which _exposed_ the problem was one which was made after reverse engineering the Apple OS, and following that more closely than Rockbox had done before. In other words a correction to Rockbox code appears to have triggered an underlying bug. That underlying bug is what needs to be fixed, not the commit which trigger
18:23:36Soaped said bug.
18:23:40roolkujhMikeS: I saw your request for an mpegplayer feature freeze. Does this include FS #7997 (make start menu optional)? This is for people who want to start playback immediately
18:24:42 Join stripwax [0] (
18:26:33KonamSoap an underlying bug? we all know apple will NEVER fix that bug just because is itching some third party app developers/users
18:26:49Konamwe have to trigger it ourselves, as I see it
18:26:57Limerrwoho, rockbox is now booting correctly, allthough i want that neat menu in the start were you get to chose if you want apple os or rockbox :S
18:27:15SoapKonam: No - an underlying bug in Rockbox. Something to do with ATA timing, If I Understand Correctly.
18:27:24LloreanThat's the current theory, yes.
18:27:56Konamoh, damn it
18:28:21SoapLimerr: that "neat menu" is the Ipod Linux Bootloader. Mind you you can install it - but we can not help you with Rockbox issues as long as you are using it - because, despite what people think, the bootloader does affect the hardware in ways which can make Rockbox behave differently than if Rockbox's bootloader was used.
18:28:43LloreanPartially reinforced by the fact that evidence at least suggests so far that it's only data read from flash while boosted that's corrupted.
18:28:59jhMikeSroolku: can it not worry about the code formatting for thumbnail right now ?
18:29:37LimerrDoesnt the rockbox bootloader have some ability to launch other os:es such as iPL or AppleOS?
18:29:45jhMikeSroolku: Patching the menu a bit won't be much bother though.
18:30:04SoapLimerr: as I mentioned earlier - it does. It can boot all three. It just doesn't have a "neat menu"
18:30:05roolkuI can remove all the stuff that just changes the code formatting
18:30:40KonamSoap so how are you dealing with that bug, we is annoying listen to music with those glitches
18:30:41 Join pixelma [0] (n=Marianne@rockbox/staff/pixelma)
18:30:55LimerrSoap, i think i can manage without the menu but how do i learn the trick to boot the others? ;)
18:31:34jhMikeSsure. what I'm doing should help keep streams as streams where they don't worry about others and make implementing lots of stuff much easier because threading control will be much better as well as taking a more object oriented approach to it.
18:31:37bluebrotherthe hold switch boots AppleOS and Play boots IPL
18:31:43SoapKonam: I am not dealing with the bug. I have little to offer as far as bug-squasing outside my standing offer to ship my iPod (free of charge) to an interested developer who has the time, talent, and motivation to look into the problem.
18:31:53roolkusounds good
18:32:41jhMikeSseeking will be controlled by a stream manager that coordinates stream threads and they simple respond to commands given.
18:32:55KonamSoap I mean, how are you dealing with it you as a user, any workaround or something
18:32:55ina84hi guys. i hv another q: why some character in text files are unreadable by the viewer plugin? it displaying a box instead. i find out the missing character is inverted commas {"} in the text. how to fix this?
18:33:36 Join Xerion [0] (
18:34:05 Quit ina84 ()
18:34:17bluebrotheruse the correct codepage setting and make sure the font you're using has the character present.
18:34:32bluebrothernot all fonts hold all characters, especially when it comes to non-latin characters
18:34:50SoapKonam: I don't use my Nano at this time. I have other players. Read the threads - people have identified the last useable daily build - I have on my webspace a series of SVN builds I did to help identify which commit triggered the issue - and the last known-working one is amongst that collection. I honestly don't recall either the address to my collection of builds, nor which version is the "last-known-good", but it should all be in the forum thread.
18:35:01 Join ina84 [0] (n=opengurl@
18:35:15 Join Morey [0] (
18:35:30bluebrotherina84: I just ansered your question ... see the logs ;-)
18:35:33roolkujhMikeS: would this be just for preview or for playback as well?
18:35:56KonamSoap can you point me to that thread
18:36:04KonamI was about to ask
18:36:35SoapKonam: Not to be a jerk, but your search button works just as well as mine. I don't recall the thread title off hand.
18:36:45 Quit andresmh (Read error: 110 (Connection timed out))
18:37:03jhMikeSroolku: for everything. threads won't terminate at the end either so it could be possible to display a menu or whatever at the end.
18:37:09Moreyroolku: I'v post a cleaner patch on FS #7997
18:37:13KonamSoap sorry, I thought you had the post at hand
18:37:19roolkuMorey: I was going to isolate the menu code from #7997 and commit it....
18:37:34roolkuMorey: great - I'll check it out
18:37:38Moreyroolku: That's what I did.
18:38:02jhMikeSroolku: they'll stay live from the so you just tell video to do a thumbnail or whatever and "seek" the stream...which is basically a reset command.
18:38:17jhMikeS*from the start
18:38:32MoreyjhMikeS: This will not effect what you are doing.
18:39:45KonamThink I found it Soap:
18:39:53 Join mf0102 [0] (n=michi@
18:40:11jhMikeSnot at all though some minor updates were done because of the object like construction. I suggest concentrating on algorthms so they can just be inserted into the stream mangement code and not depend on a particular structure.
18:40:15roolkujhMikeS/Morey: the only thing I am not sure about are the optimisations in video_out_rockbox.c (at the end of
18:41:00 Join kugel|afk [0] (i=kugel@unaffiliated/kugel)
18:41:00 Quit kugel (Read error: 104 (Connection reset by peer))
18:41:07 Nick kugel|afk is now known as kugel (i=kugel@unaffiliated/kugel)
18:41:10roolku(as far as impacting on jhMike's stuff is concerned)
18:41:29MoreyThey won't effect it.
18:41:34 Quit stripwax ("Miranda IM! Smaller, Faster, Easier.")
18:41:46jhMikeSroolku: are you just displaying full size now?
18:41:47 Nick advcomp2019_ is now known as advcomp2019 (n=advcomp2@
18:41:58jhMikeSoh, svm
18:42:42jhMikeSI'll probably get a couple conflicts but easily resolvable
18:43:57Moreyroolku/jhMikeS: They are just algorithm simplifications. One line for one line, shouldn't even offset the file.
18:44:58roolkuokay, done :)
18:45:11jhMikeSI did add "mode" to vo - thumbnail or fullsize. Basically calling vo_draw_frame via a pointer.
18:45:53roolkujhMikeS: full screen won't have a progress bar though, I assume?
18:46:08MoreyjhMikeS: You mean for fullsize thumbnail or compining draw_frame and draw_thumb
18:46:10jhMikeSnot yet
18:46:21 Join stevenm [0] (n=stevenm@
18:46:23Moreycompining = combining
18:46:43J3TC-Crap...I can't apply both the progresslider bar batch and album art patch
18:46:55stevenmn1s, Hello. I think there may be a bug with the new MIDI loop.
18:46:56jhMikeSMorey: no, selecting the function to call when setting the mode...vo_draw_frame_full or vo_draw_frame_thumb
18:47:30roolkujhMikeS: the other thing to watch out for is the different buttons when the thumb and full screen are 90 degrees rotated
18:47:31stevenmn1s, I've got a test file right here: stevenm/1.mid">
18:47:47jhMikeSyou just send a message to video to draw a thumbnail
18:47:54stevenmn1s, something ends up corrupting memory and overwriting a struct, which leads to a major crash.
18:48:14MoreyjhMikeS: That sounds good.
18:48:38stevenmn1s, This happens because the file tries to change its tempo on the fly. That causes numberOfSamples to change because of tick()
18:48:44jhMikeSroolku: no button changes are should draw as-is
18:49:37jhMikeSui and buffering will have their own threads...ui on the main thread, buffering on it's own
18:50:26MoreyjhMikeS: Does that mean the user can be interacting while buffering in the background? This would speed things up.
18:50:37roolkuthe button control could get confusing, but lets see how it turns out
18:51:10jhMikeSMorey: sure and you'll be able to view the menu from paused which doesn't seem to work right now
18:51:49jhMikeSroolku: it's _way_ simpler. the UI stuff just makes calls like mpeg_stream_pause/play/stop/seek.
18:51:54MoreyjhMikeS: Nice. Lots a luck! :)
18:52:24jhMikeSpretty close to done with it actually. I just need the buffering straight and should be able to start testing.
18:52:44stevenmn1s, also, because numberOfSamples changes, it may be affecting the foor loop condition in midilay.c (the one taht calls synthSamples)
18:54:19 Nick Arathis is now known as Arathis|busy (
18:54:24 Quit Morey ("Ninja IRC v1.5.8.1(#1) exiting after 19m26s of use")
18:55:29 Join mrkiko [0] (
18:55:39mrkikoHi all! I'm finally back
18:56:22n1sstevenm: I know, about this issue but have not had any crashes yet, however one file gets stuck in an endless loop and never finishes, I'm thinking about a good way to fix it...
18:57:04stevenmn1s, Ah, all right. I was up til I think 6AM trying to hunt the bug. Just woke up and there it was
18:57:15n1sstevenm: btw I noticed another problem that I think was caused by the pitch bend depth, one of my files has high pitched squeaks now...
18:57:32 Quit FOAD (Remote closed the connection)
18:57:41stevenmn1s, that file I posted does it. I deleted the measures so it does the crash about 3 seconds after you start playing it
18:58:00 Join davina [0] (
18:58:15stevenmn1s, numberOfSamples goes from 100something all the way up to 260, then suddenly to 95
18:58:56stevenmn1s, due to tempo events. I guess the logic can't handle that... also, the sample buffer within synth.c is set to 256 max, and then num_samples is multiplied by 4
18:59:37 Join stripwax [0] (
19:00:18n1sstevenm: I'll look into it more, kind of busy now, I have an exam on tuesday :-/
19:00:24stevenmn1s, when it writes past buffer, it overwrites some structures.. like the pointer to the midi file itself.
19:00:44stevenmn1s, oh, that's fine... i'm just saying the results of what I found. It took a long time
19:01:11stevenmn1s, last night I ported over the whole plugin to be a standalone application just so I could use GDB watchpoints, which apparently don't work in plugins
19:01:30n1shere is one file that now plays with squeaks I tested with revisions prior to the loop/pith bend commit and it was fine and then patched with the loop change and it still played fine so i think it is caused by pitch bend depth
19:01:43stevenmn1s, and when I finally set up the watchpoint, memory alignment was such that it was overwriting a DIFFERENT structure.. then gdb hung.. it sucked :(
19:02:08stevenmn1s, Thanks, will look. It could also be the patch 'prescaler' code also
19:02:25n1sstevenm: it occured before that commit
19:02:36stevenmn1s, aah, ok.
19:03:11n1ssorry that bug has caused you so much trouble :-/
19:04:06stevenmn1s, s'alright..
19:04:07stevenmn1s, you are referring to the beeps that happen randomly in the music?
19:04:39stevenmthere it is... yea, somewhere delta is messed up. printf time
19:05:51stevenmunless something else is messing up those arrays too.. but i hope not
19:07:34J3TC-For the patches that has a linethrough them
19:07:39J3TC-Does that mean they are already implemented?
19:10:45mrkikoSomeone knowing something about bug 8003? Some testing needed? Some progress? I want be cooperative as much as possible!
19:12:56stevenmn1s, Yeah, it's gotta be the pitch bend. Note offsets are something like 63.
19:13:21stevenmn1s, the >> operation. Is it a logical shift or arithmetic shift?
19:14:06bluebrotherafaik it's logical
19:14:11 Quit stripwax ("Miranda IM! Smaller, Faster, Easier.")
19:14:16 Join FOAD [0] (
19:14:19***Saving seen data "./dancer.seen"
19:14:37stevenmthat might explain it... maybe..
19:15:17bertrik>> depends on the signedness of the variable that is operates IIRC
19:15:41 Join ilgufo [0] (
19:15:49 Quit Limerr ()
19:16:13stevenmor not
19:18:02jhMikeSbertrik: if x is signed x >> n is an arithmetic shift and fills the sign bit from the left, otherwise it's a logical shift and x is zero-filled from the left.
19:18:10stevenmn1s, do you know anything about the pitch bend depth?
19:18:41 Join andresmh [0] (
19:19:01bertrikjhMikeS: yes, that's what i remember, looking for confirmation right now
19:19:10stevenmn1s, I think the issue is decoding the bend depth event. I found online that it is just the Data Ent MSB (LSB?) event, but I think that controller has multiple functions based on some other bits
19:19:24jhMikeS>> also rounds toward -inf whereas / rounds toward 0.
19:19:58 Quit andresmh (Client Quit)
19:20:10jhMikeS(-1) >> 1 = -1, (-1)/2 = 0
19:21:35 Join stripwax [0] (
19:22:05n1sstevenm: I'm not familar with that stuff at all
19:22:26bertrikhmm, interesting
19:22:39stevenmyes, that's what it is. The file tries to set controller 6 to things like 75 and 84, which is WAAY too much of a bend. The math is right, but the sequencer is treating non-depth events as depth events
19:23:17 Quit Rob222241 ()
19:23:29stevenmRight now it just takes any Data Ent MSB (6) write and treats it as bend depth. I'm looking at the data now.. there appear to be other events around it that tell it what that write should do. And midiplay doesn't look at those yt
19:24:21stevenmI need to go eat and do homework, but I can google around for a better description of this stuff later and hopefully put in a fix
19:28:09 Quit ashes ("leaving")
19:28:44 Join sounddude [0] (
19:29:36mrkikostevenm: are you a student?
19:29:57stevenmmrkiko, yep. senior undergrad
19:31:01J3TC-How do you sync patches to svn?
19:31:11J3TC-A lot of patches break just even a week or two old :3
19:32:13n1sJ3TC-: you open the patched file and the .rej file and see which hunks didn't go in and put them in, in the correct place with additional adjustments as needed
19:32:34J3TC-Ah ok
19:32:52n1ssometimes it will take actual programming effort but mostly it's cut 'n past
19:34:15J3TC-So it's just the lines are off then
19:34:28stevenmn1s, ok, found the information. I'v got a fix in mind, but will code it up once i go get some food. See ya
19:34:45stevenmdoesn't look bad, just another reg value to track
19:35:03n1sgreat, love the improvements lately :-)
19:36:07 Join Rob2222 [0] (
19:36:21n1sI think the out of bounds memory writing is the "only" problem both your file and mine that doesn't crash but never ends does it...
19:36:28 Quit mozuch (K-lined)
19:37:47n1sbtw is there a defined max value of the number of samples in a tick?
19:40:24J3TC-- result = rb->menu_show(m);
19:40:41J3TC-The ones with the "-" you remove from the code?
19:41:00J3TC-Thanks :3
19:41:34 Join XavierGr [0] (
19:43:14 Join Rob222241 [0] (
19:47:39 Quit FOAD ("I'll be back")
19:47:49 Quit Llorean ("Leaving.")
19:49:32 Quit MethoS- (Read error: 101 (Network is unreachable))
19:49:47 Join FOAD [0] (
19:50:40 Quit amiconn (Nick collision from services.)
19:50:47 Join amiconn [0] (n=jens@rockbox/developer/amiconn)
19:51:14*J3TC- does a test build
19:51:17*J3TC- crosses finger
19:53:07 Quit GodEater_ (Read error: 110 (Connection timed out))
19:53:08 Quit moos ("Glory to Rockbox")
19:56:29 Join Calcipher [0] (
19:57:51 Quit ilgufo (Nick collision from services.)
19:57:54 Join gufo [0] (
19:59:55 Quit Rob2222 (Read error: 110 (Connection timed out))
20:03:40 Join Domonoky_ [0] (
20:04:03 Quit FOAD (Remote closed the connection)
20:06:18 Join FOAD [0] (
20:07:11 Join mo-seph_ [0] (
20:07:26 Quit spiorf (Connection timed out)
20:07:49 Quit mo-seph_ (Remote closed the connection)
20:07:53 Quit bertrik ("bye")
20:08:15 Join spiorf [0] (
20:11:14 Quit ina84 (Read error: 104 (Connection reset by peer))
20:12:36 Join ina84 [0] (n=opengurl@
20:13:01 Join DrNI [0] (
20:13:24 Join webguest72 [0] (i=4ae94966@gateway/web/cgi-irc/
20:13:34DrNIanybody out there?
20:13:41DrNII just installed rockbox
20:13:50DrNIon an ipod video 5.5g
20:14:03DrNIhowever, I can't exactly find out how to get themes to work
20:14:09n1swell, congratz
20:14:14 Quit Konam (Read error: 110 (Connection timed out))
20:14:22webguest72I have it installed on a Sansa E270 and a Gigabeat F40
20:14:33DrNIit's my girlfriend's ipod, so it must be good looking, you know... woman's acceptance factor, and stuff
20:14:38n1syou unzip them to your ipod and select them from the browse themes menu
20:14:49DrNIthat's what I did
20:14:57DrNIand it picked a nicer font and color
20:14:59DrNIbut not the icons
20:15:01bluebrotheror you use rockbox utility for theme installation
20:15:10webguest72You can download themes from the rockbox theme page, unzip them to your players root directory, then use browse themes to select it
20:15:28 Join DataGhost [0] (
20:15:47DrNIwebguest72: root is / not /.rockbox/ ?
20:16:04n1sDrNI: were the icons included in the zip? root is /
20:16:17DrNIwait a sec
20:16:33 Quit FOAD (Remote closed the connection)
20:16:34webguest72root is /, I think the zip files containing the themes are based on install in root, and include .rockbox in the path
20:16:35 Join Konam [0] (
20:16:46n1sso a wps file should end up in /.rockbox/wps for example
20:16:53DrNIhm yeah so I did right
20:16:59DrNII tried the
20:17:07DrNII put everything in place in .rockbox
20:17:18n1swere there icons in the zip?
20:17:38DrNIthey are in /.rockbox/wps/iRocker
20:17:56n1sthat's wps images, not icons for lists/menus
20:18:29DrNIyeah but the player doesnt show images either
20:18:31 Join toffe82 [0] (
20:18:49n1sDrNI: so your wps looks as the default?
20:18:50DrNII mean, the player application, or window, or what you call it :)
20:19:04DrNIyes, everything but the font looks like default
20:19:11n1swps is what it's called, where did you get the zip
20:19:15DrNIand the background image, which is I guess defined in the theme file
20:19:28Calciphermy sansa says 74% 3h 49m under system/rb info
20:20:15DrNI <- I used this one
20:20:44 Quit mo-seph (Read error: 110 (Connection timed out))
20:20:49DrNIso rockbox seems to regard the stuff from the theme file
20:20:55DrNIbut not so the things in the wps
20:21:30Calcipherwow only minutes later my player says battery is 71% 3h 40m
20:21:42n1sDrNI: that's because the wps file either requires patches not found in the official build or is broken
20:21:55DrNII used the current built
20:22:01n1sCalcipher: it's not been calibrated, the time means nothing
20:22:03 Quit Domonoky (Read error: 110 (Connection timed out))
20:22:27DrNInls: it came with some themes already
20:22:32n1sDrNI: yes and that doesn't include every hacky wps patch out there that some of the themes require
20:22:38DrNInls: e.g. there is a iCatcher theme included
20:22:43DrNIbut this doesn't show icons either
20:23:01n1sDrNI: which icons are you talking about?
20:23:13DrNIin menus, in the player view
20:23:31n1sDrNI: that's because most themes don't include them
20:23:55n1sDrNI: check out this page
20:24:05DrNIah ok the iCatcher loads the images in the player view
20:26:41 Join FOAD [0] (
20:27:55DrNIhm I guess the thing is that I don't have support for showing the cover images
20:28:00DrNIwhich the iRocker theme wants to do
20:29:11 Quit FOAD (Client Quit)
20:29:23*mrkiko builds the SIM
20:31:50mrkikoDoes the voice system work on the SIM?
20:33:37 Join RoC_MasterMind [0] (
20:34:52 Part pixelma
20:36:05mrkikon1s: thank you
20:36:31 Join FOAD [0] (
20:36:38 Quit amiconn (" Fensteraufdatum")
20:37:38mrkikoAnd when the SIM starts, where does it pint oon my filesystem?
20:37:39jhMikeSugh, I can't browse files now while database is updating. It just hangs until finished. other simultaneous file access is fine.
20:37:45mrkikoarchos/ dir?
20:37:50*bluebrother returns and wonders what protocol bluehttp: is ...
20:38:11bluebrothermrkiko: yes.
20:38:16 Join AmbiquitY [0] (i=51f1b7d9@gateway/web/cgi-irc/
20:38:23J3TC-Can you do copy/paste in cygwin?
20:39:06 Quit newbyx86 ()
20:39:12 Quit FOAD (Remote closed the connection)
20:39:34mrkikoNOTE: while zipping a sim-build, you forget to inclue the "rockboxui" executable.
20:39:41AmbiquitYhi everybody, I've got a problem on my ipod nano.. it doesnt boot, i have a blank screen and doesnt do anything. it doesnt react to a hard reset (menu+select) not the usb cable.. nothing what can i do?
20:39:52mrkikoSorry: guidelines says to not misuse capital letters...
20:40:17stripwaxAmbiquitY : - maybe the battery is flat?
20:40:28stripwaxHave you tried charging it fully from a mains (wall socket) charger?
20:40:43bluebrotherJ3TC-: try clicking the icon on the top left of the window
20:40:46 Join kratonator [0] (
20:40:53bluebrotheriirc this has an entry for cnp
20:41:10AmbiquitYi dont think the battery is flat, and i dont have a wall socket charger :s
20:41:17kratonatorhi all
20:41:37 Join newbyx86 [0] (
20:41:38stripwaxAmbiquitY - if it doesn't turn on, and doesn't react to hard reset, either the battery is flat or it is broken
20:41:51AmbiquitYwouldnt a wall socket thing have the same effect as plugging it in my computer?
20:42:10kratonatormight be old news but I've heard that the new Sansa View has a GoForce 6100
20:42:14stripwaxyes and no. usb delivers lower current
20:42:34kratonatorit says that it has ARM archtecture
20:43:24kratonatoris there a chance to port rockbox on it once it's out?
20:43:41AmbiquitYthis happend when i wanted to plug in my ipod but i didnt want the black and white data transfer thing and i forgot to put on hold on time, so i just pulled out the cable out of my ipod.. could this be the reason that its not doing anything?
20:43:43stripwaxAmbiquitY - according to apple, the ipod might not respond for 30 minutes after plugging into computer usb if battery is flat.
20:44:04AmbiquitYok I'll leave it plugged in for some time and hope for the best
20:44:06 Join ]RowaN[ [0] (
20:44:06stripwaxAmbiquitY - the black and white data transfer thing.. ??
20:44:16AmbiquitYyes i forgot the name
20:44:24AmbiquitYwhen u get the do not disconnect screen
20:44:24stripwaxi don't know what it is .. ?
20:44:28stripwaxoh right.
20:44:40AmbiquitYand uhm.. i disconnected
20:44:42]RowaN[guys can you give me write permissions for the wiki please? my wiki name is RowanPestille
20:44:44stripwaxyou must always click on the 'safely disconnect' otherwise the filesystem could get corrupted
20:44:45krazykitkratonator, there's a chance for anything, really, it's just a matter of if there are people who have the device and the ability to get code running.
20:44:50stripwaxif you did that all should be fine
20:44:51maxkelley"comes equipped with all the soup-to-nuts features typically found in the Sansa audio line."
20:45:01AmbiquitYbut i didnt,
20:45:01maxkelleyI've never heard of that term befoe.
20:45:19AmbiquitYbut i did it alot before and never had problems.;
20:45:24stripwaxAmbiquity - it'll probably still be ok after a full charge, maybe it didn't turn off properly and the battery drained
20:45:42 Quit ina84 ()
20:45:50n1swow that goforce thingy features "iPod quality audio playback"
20:46:01kratonatorwell the second gen nano is pretty popular (people have the device) but it is still impossible to get any code running
20:46:11AmbiquitYok I'm not touching it anymore I'm going to watch a movie and just leave it plugged in, thanks for your help
20:46:27kratonatori know, i know...but i hope nvidia wasnt serious when they said that
20:46:43krazykitkratonator, not impossible, it's just that no one has figured out how to run non-apple code on it.
20:46:46stripwaxok. if it still doesn't work, time for apple support..
20:47:28kratonatorkrazykit, ok, ok....i get it
20:47:44 Quit stripwax ("Miranda IM! Smaller, Faster, Easier.")
20:48:19AmbiquitYwill my warranty be lost if they find out rockbox was installed?
20:49:51 Quit ompaul (Client Quit)
20:50:32 Join ompaul [0] (n=ompaul@gnewsense/friend/ompaul)
20:50:44 Join FOAD [0] (
20:51:47 Quit AmbiquitY ("CGI:IRC (EOF)")
20:52:24 Quit mrkiko ("leaving")
20:53:26 Join mrkiko [0] (
20:53:35mrkiko... sdl has messed up my screen
20:53:51 Quit gufo ("So Long, and Thanks For All the Fish -")
20:54:05mrkikoSo I resized correctly some terminals and now I'm rocking again. Aniway, I can't get the audio working on rockbox, even if the audio works well in my system.
20:55:33DrNIok I now used the simplex theme which looks nice (she thinks it does ;)
20:55:48DrNIand is supported by the current built for this ipod model
20:55:53DrNIthanks for your support
21:00:00 Quit FOAD (Remote closed the connection)
21:02:08 Join FOAD [0] (
21:03:10 Quit Konam (Remote closed the connection)
21:03:45 Join feindbild [0] (
21:03:48feindbildhi =)
21:06:09 Join Jeton [0] (n=chatzill@
21:06:53mrkikoCan someone help me with the volume on the simulator can be trashed up with the "+" key?
21:07:04feindbildis anybody using rbutilqt successfully? I always get errors like: Object::connect: No such signal HttpGet::downloadDone(int,bool). I thought it was linked statically against Qt4?
21:07:40rasherfeindbild: it should be. Which platform is this?
21:08:04feindbildrasher: Linux / Slackware -current (only Qt3 installed)
21:10:55 Quit DrNI ("Leaving")
21:11:00webguest72Question about the database browser... on the gigabeat F port... when I drill down to a selection of tracks and pick one with the right arrow the first track
21:11:41webguest72on the list starts playing, not the one I picked. I can use the right arrow in the WPS to get to the track I want. What's wrong here?
21:12:00mrkikoWho can help me with the simulator? I can't get the audio working: i have the needed sdl components and I get no errors but looking at the debug screen
21:12:15mrkikoI can see rockbox opens a ny mp3 files, analyze it and closes it without actually plaing.
21:12:32 Join andresmh [0] (
21:12:34mrkikoIt is using mpa codec for mp3 files...
21:12:34rasherwebguest72: I think there's a setting "Play selected first", which will play the one you selected first, rather than start from the top of the list, but I'm not sure that option does this, or what it does. I'm sort of guessing here
21:13:04andresmhis there a way to play in shuffle mode all the files in a directory from the file browser?
21:13:24markunandresmh: you can do "insert shuffled" on the dir
21:14:15 Nick idnar_ is now known as idnar (i=mithrand@unaffiliated/idnar)
21:14:21andresmhmarkun: so i am looking at a directory, i click select and i get the context menu, should i see "insert suffled" there?
21:14:24***Saving seen data "./dancer.seen"
21:14:40webguest72Yup, play selected first fixes that. Thanks!
21:14:54n1sandresmh: yes you can select it in the context menu
21:15:09rasherwebguest72: I believe that option is on by default, so you probably changed it without knowing what it did
21:15:10n1sor you can turn on shuffle and just start playing a file in the dir
21:15:25markunandresmh: I never use shuffle myself, perhaps there are other ways to do it
21:15:38andresmhit's just that my folder has subfloders
21:15:58andresmhand in the context menu i don't see an "insert suffled" option
21:16:03webguest72Another thing on my gigabeat was that after connecting the usb (while in rockbox) to download files, then disconnecting it
21:16:17n1sandresmh: you can still insert the dir into a playlist, by default all subdirs are inserted too
21:16:37mrkikoDoes the wiki contains some hintgs about the sim' I can't find them...
21:16:40markunn1s: are you sure it's on by default?
21:16:40webguest72and rebooting to rebuild the database my clock setting was gone! (had to reset the time). Also if I plug in the usb while the unit is off
21:16:56n1smarkun: yup, I changed it about a week asgo :-)
21:17:00andresmhoh i found the insert shuffled option now
21:17:09webguest72it will connect to the pc but gets locked up. I had to cycle the battery to recover. connecting while in rockbox avoids this.
21:17:13andresmhwill it create a new playlist?
21:17:33markunandresmh: if you stop playback first it will create a new playlist
21:17:38n1sandresmh: not if you have one currently active (== not stopped)
21:18:42andresmhi am getting No file! errors
21:19:43andresmhcould it because some of my filenames are non-english chars?
21:21:05andresmhI get this when I click on view current playlist "error accesing playlist control file"
21:21:47webguest72is there a function in the file menu that will give the remaining space on the disk. Or do I have to mount the player and check this from the pc?
21:22:01mrkikorockbox info
21:22:12andresmhand when I select Playlist Catalog I get "No playlists"
21:22:13mrkikowebguest72: information -> rockbox info
21:22:27feindbildsummer of the code is over, right?
21:22:54feindbildhow far did the USB project (sansa e200) come? *g*
21:23:12webguest72Thanks I found it. There's lots of stuff burried in there!
21:23:48n1s-> <- this far, there is different usb code in development which has gotten further though
21:24:13 Quit Lear ("ChatZilla [Firefox]")
21:24:14feindbildn1s: =)
21:24:50*feindbild is quite happy with rockbox =D
21:25:02andresmhI keep getting "error accessing playlist control file", is there a way i can clear my playlists and start fresh?
21:25:20andresmhi already rebooted
21:25:25andresmhand also cleared settings
21:25:51andresmhI just installed rockbox a few minutes ago and loaded 10GB of files into the ipod
21:26:29andresmhI have r15249
21:28:38J3TC-Anyone know anything about the scrolling margins patch?
21:28:55J3TC-It seems to be out sync and is more than just lines being off
21:29:18J3TC-The patch fails here
21:29:22J3TC-#if defined(HAVE_LCD_BITMAP)
21:29:22J3TC-- screens[i].setmargins(0, 0);
21:29:22DBUGEnqueued KICK J3TC-
21:29:22J3TC-+ screens[i].setmargins(0, screens[i].width, 0);
21:29:47 Join stripwax [0] (
21:29:47J3TC-at apps/menus/main_menu.c
21:29:53SoapJ3TC-: stop spamming the channel
21:30:01Soapuse pastebin for multiline pastes.
21:30:12J3TC-And I looked at it and it doesn't have #if defined(HAVE_LCD_BITMAP)
21:30:15J3TC-Don't know what that is
21:30:19J3TC-and stop being anal
21:30:52 Join keanu|away [0] (n=chatzill@unaffiliated/keanu)
21:31:15rasherJ3TC-: Please respect the rules we have in this channel.
21:31:23 Quit webguest72 ("CGI:IRC (EOF)")
21:31:49J3TC-I apologize for spamming 5 lines
21:32:34 Quit andresmh ("ChatZilla [Firefox]")
21:32:42rasherI don't see why you think this is something to be upset about? It's clearly stated in the rules, and they're there for a reason. If you don't like that, you're free to not use this channel
21:33:26J3TC-It's 5 lines which I seriously doubt count as spam
21:33:43J3TC-But if you feel that it is then so be it.
21:33:47J3TC-I'm not going to argue about this.
21:33:52J3TC-I was just asking a question
21:33:54Soapplease read and respect the channel guidelines. On top of that the scrolling margins patch is unofficial and not appropriate subject for conversation in #rockbox. The patch tracker is really a better place. - That being said it is also considered clutter to ask for a sync in the patch tracker - as those who are unwilling/unable to sync are really at the mercy of those who are able/willing. The polite thing to do is sit on your hands and wait for the k
21:33:54Soapindness of others, or to work on syncing the patch yourself. ALL that being said - such rejected features are really a grey area in the #rockbox / #rockbox-community division and perhaps talking about it in the latter (as long as you can keep off topic) is better.
21:33:56rasherSounds swell.
21:33:57keanu|awayJ3TC-: for some it can be considered spam
21:34:15keanu|awayJ3TC-: I also have the same question about scrolling margins - you can see the changes in SVN
21:34:31SoapJ3TC-: your constant useage of the "Enter" key as if it were a comma is also spam. How about starting and completing one sentance on one line?
21:35:15J3TC-Wow soap....way to go push this so far
21:35:17J3TC-I said I got it
21:35:45 Quit davina (Remote closed the connection)
21:36:59 Join ilgufo [0] (
21:38:13 Join Tavnos [0] (
21:38:27TavnosHello Nico_P
21:40:10 Join davina [0] (
21:41:37 Quit davina (Client Quit)
21:45:35 Quit Gnu47 (Excess Flood)
21:45:48 Join Gnu47 [0] (
21:45:48 Join salty-horse [0] (n=ori@pdpc/supporter/active/salty-horse)
21:46:37 Join davina [0] (
21:48:07salty-horsedoes the "(B)ootloader" option in tools/configure mean that the bootloader is built in addition to the (N)ormal setting?
21:48:25krazykitsalty-horse, no, it only builds the bootloader
21:48:36stevenmn1s, I committed a fix to the pitch bend problem.
21:49:11stevenmThat isn't all though... Apparently there are other events that can set pitch bend depth, but I haven't seen a file that uses them yet.
21:49:15salty-horsehmm.. more to the point: how can I build the sansapatcher? it seems the bootloader option isn't it :)
21:49:38stevenmmost that I've seen use controller 100 and 101 set to 0, then a write to controller 6 to set the actual depth
21:49:38kugelinstaller option?
21:50:25kugeli'm not sure though
21:50:30 Quit davina (Client Quit)
21:50:39kugelyou can trunk sansapatcher source seperately
21:51:22mrkikoNo... I cant get the sim audio working...
21:51:27Bagderme and Linus are now back on Swedish soil
21:51:34mrkikoIs someone familiar with it?
21:51:43salty-horsekugel, I think already have it in rbutil/sansapatcher, but make says " No rule to make target `firmware.mi4', needed by `bootimg_c200.c'."
21:51:45rasherBagder: brought home any cheap counterfeit electronics?
21:51:59SoapSo...the question of the hour is...^ rasher beat me to it.
21:52:06stevenmlike a fake ipod shuffle
21:52:08Bagderrasher: no, I didn't end up finding anything interesting worth spending my money on
21:52:20kugelsalty-horse: you need to download the bootloaders of e200 and c200
21:52:23Bagderfake ipods were everywhere
21:52:30stevenmBagder, where did you go?
21:52:38BagderBejing, China
21:52:44kugelor build ;)
21:52:50salty-horsethe rbutil thing is just a wrapper?
21:52:53stevenmBagder, awesome! my buddy went there. so much stuff
21:53:22 Quit mf0102 ("Verlassend")
21:53:23BagderI'll need to go sleep now, I'm jetlagged like crazy
21:53:41 Quit Gnu47 (Excess Flood)
21:53:43stevenmmy friend got an ipod shuffle from there. looked fine, worked, until itunes didn';t see it as an authentic device
21:53:51 Join Gnu47 [0] (
21:53:58salty-horsekugel, I'm confused - the manual says the sansapatcher source is indeed in "rbutil/sansapatcher/" - how do I build it?
21:54:08 Join davina [0] (
21:54:21kugelyou need to copy the bootloaders of c200 and e200 in that dir
21:54:33salty-horseah :)
21:55:02 Quit davina (Client Quit)
21:56:06 Join grndslm [0] (
21:56:19salty-horsekugel, when building the (B)ootloader, which files do I need from the resulting dir?
21:56:55kugelfirmware.mi4 for c200 and PP5022.mi4 for e200
21:57:16stevenmwell, i'm out. bye guts
21:57:22 Quit stevenm ("Connection reset by beer")
21:57:32 Join davina [0] (
21:58:15 Quit davina (Remote closed the connection)
21:59:10 Quit Jeton (Read error: 110 (Connection timed out))
21:59:28 Join kubiix [0] (
21:59:52 Quit miepchen^schlaf (Read error: 110 (Connection timed out))
22:00:08 Join miepchen^schlaf [0] (
22:00:21 Join davina [0] (
22:01:44salty-horsekugel, I can find PP5022.mi4 but no firmware.mi4
22:02:03kugeldid you build the bootloader for c200?
22:02:09kugelyou need to do that too
22:02:30krazykiti thought sansapatcher could use an external bootloader by a command line switch
22:02:31salty-horseI chose "e200" from the configure menu
22:03:04salty-horsekrazykit, I'm just trying to build it. I don't have sansapatcher yet to give it any switches
22:03:32krazykitsalty-horse, right, i mean you could download a prebuilt one and just do sansapatcher -a yourbootloader.mi4
22:03:33Domonoky_salty-horse: if you want to build sansa patcher without builtin bootloaders.. define RBUTIL :-)
22:03:50 Join desowin [0] (
22:04:29salty-horseare the pre-built ones up to date? specifically ?
22:04:53 Quit mrkiko (Remote closed the connection)
22:05:03krazykitsalty-horse, they all should be, yes
22:05:05kugelsalty-horse: you can just build the c200 bootloader
22:05:11salty-horse(I want the new feature from r15225
22:05:49kugeljust repeat the actions for the e200 bootloader, but choose c200
22:06:37salty-horsegetting a 404 on
22:06:44salty-horsekugel, are you sure it won't break anything? :D
22:07:02kugelhow should that break something?
22:07:21salty-horseI'm not the rockbox expert :)
22:07:29 Nick parafin is now known as parafin|away (
22:07:47kugeljust build the bootloader for c200, i don't see why that should be a problem
22:08:25rashersalty-horse: to build sansapatcher you need both bootloaders
22:08:38 Quit davina (Remote closed the connection)
22:08:54 Quit ]RowaN[ ()
22:09:23salty-horserasher, are there are written instructions on how to do that? I can't seem to build the required firmware.mi4
22:09:41rasherNo idea, to be honest. Never did it myself
22:10:35 Join davina [0] (
22:11:06 Quit davina (Client Quit)
22:12:31 Join davina [0] (
22:14:38 Join Zagor [0] (
22:16:48kkurbjunjhMikeS: do you know what changes would need to be made to use lcd-as-memframe for the m:robe also?
22:17:21kkurbjunI am getting errors like this when I try and compile it for the mrobe: target/arm/lcd-as-memframe.S:191: Error: address offset too large −− `strh r1,[r0],#480'
22:20:23salty-horsekugel, rasher, I finally understood what you said about sansapatcher needing both firmwares −− a bit inconvenient but it worked! Thanks, Barry, for this great new feature of disabling database rebuild!
22:20:59 Quit stripwax ("Miranda IM! Smaller, Faster, Easier.")
22:21:43 Quit ilgufo ("So Long, and Thanks For All the Fish -")
22:23:33 Quit jhMikeS (Nick collision from services.)
22:23:36 Join jhMikeS [0] (n=jethead7@rockbox/developer/jhMikeS)
22:24:26kkurbjunjhMikeS: did you get that last message?
22:26:43jhMikeSkkurbjun: yeah
22:27:03jhMikeSI guess the increment is too big there
22:28:33jhMikeSI'd expect it to complain about more lines than that I think
22:28:50kkurbjunthere are a couple of them
22:29:08 Join Fraser [0] (
22:29:22jhMikeSis m:robe portait or landscape?
22:29:36kkurbjunline 191, 220, 249, 408,454, and 502
22:29:47kkurbjunit's portrait (480x640)
22:30:11scorche(other way around)
22:30:24scorcheerrr....kidding >_>
22:30:38CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
22:30:38*scorche hunts for breakfast
22:30:53 Join pixelma [0] (i=pixelma@rockbox/staff/pixelma)
22:31:55pixelmasalty-horse: I read you want the bootloader for c200 to stop the database refresh? Doesn't work for it as the commit message states
22:32:03scorchekkurbjun: wait...i thought iot was landscape
22:32:32jhMikeSthey should share the same file then. The displacements are limited but some #if blocks can be used
22:33:04salty-horsepixelma, I have an e200 - I just had to compile the c200 bootloader in order to compile sansapatcher
22:33:17 Quit sneakums (Remote closed the connection)
22:33:20pixelmaI see, misunderstood then :)
22:34:05kkurbjunscorche, it is landscape, but teh hardware is setup for portrait, so that's what we're working with right now
22:35:20 Quit keanu|away (Read error: 104 (Connection reset by peer))
22:35:58 Join sneakums [0] (
22:37:10 Quit Fraser (Read error: 104 (Connection reset by peer))
22:37:25 Quit merbanan ("Leaving")
22:37:36kkurbjunjhMikeS: so it's failing because an immediate value can only be 8 bits for this function?
22:40:04jhMikeS480 is 4 bits, but the displacement can't be that large
22:40:51kkurbjunbut it could be written as 1E, E?
22:41:44kkurbjundoes the assebler take rotations into account automatically?
22:42:20jhMikeSyes. I'll sec
22:43:19kkurbjunjhMikeS: what's the problem with the code if the screen is landscape?
22:43:33kkurbjunwell, actually, it doesn't matter on the mrobe..
22:43:46 Join linuxstb [0] (
22:43:52kkurbjunit would all have to be rewritten since the hardware won't do it for you
22:45:02jhMikeSif it's landscape, then landscape YUV routines would be needed. the ones there rotate by 90deg.
22:46:15jhMikeSHope I set it up right. The wider screen needs an extra add instruction where the smaller ones can use displacement.
22:46:27 Join amiconn [0] (n=jens@rockbox/developer/amiconn)
22:46:52 Quit Frazz (Read error: 110 (Connection timed out))
22:48:37kkurbjunnice, I don't have any videos to test with at the moment, but once we're ready for it I'll test it then
22:49:02linuxstbkkurbjun: How fast is your CPU?
22:49:11jhMikeSIt's should be the way I had it before I removed that extra add
22:49:34kkurbjun~180 MHz for the ARM core
22:49:48kkurbjunI think the dsp will be needed to get descent fram rates
22:50:01linuxstbAt 640x480, definitely...
22:50:11linuxstbWhat can the original firmware decode?
22:50:38*jhMikeS still wonders how to hide the DSP behind libmpeg2 calls...would be rather strange
22:51:01kkurbjunthe original firmware doesn't have any video decoding
22:51:15kkurbjunthey have a TV out for it though
22:51:30kkurbjunI think those neuros set top devices are based on the same chipset though
22:51:35kkurbjunand I think they do mpeg4
22:51:37linuxstbPresumably (being Olympus) it's designed more for photo viewing?
22:52:06kkurbjunyes, it has a great display for photos, and it looks very nice on the TV as well
22:52:21linuxstb4 times the number of pixels as the Gigabeat F, and a slower CPU..
22:52:34kkurbjunthere's a limited angle that you can see well though
22:52:59kkurbjun:), yeah, but it has a DSP too.. too bad it's all closed source tools to actually use it
22:55:08Buscheljust submitted the yuv-blit for iPod Video using assembler and chroma buffer, gives a nice speed-up :)
22:56:05linuxstbNice. How much?
22:57:28Buschelroundabout +20% vs. trunk
22:57:40linuxstbSo what FPS before and after?
22:57:57Buschelbefore 7.2, now 8.7 at 30MHz
22:58:41 Quit XavierGr (Nick collision from services.)
22:58:44 Join XavierGr [0] (
22:59:43Buschelfs #7951
22:59:47 Quit Tavnos ()
22:59:51linuxstbWhat about at 80MHz?
23:00:27linuxstbAh, you're just talking FPS for the lcd blit?
23:02:14amiconnHmm, less than what I thought should be possible with asm
23:02:54 Quit sneakums (Remote closed the connection)
23:02:58Buschelstill the problem with 14ms delay when waiting for the LCD bridge for each frame
23:03:25amiconnThat delay imposes a theoretical upper limit of 70fps...
23:04:13 Join now [0] (
23:04:20Buschelnevertheless you may just skip this wait for fps measurements :o)
23:04:23nowany chance porting iaudio g3?
23:04:40Buschelbut you won't gain any fps when playing videos
23:04:41nowsomeone tried workin on it yet?
23:04:53 Join sneakums [0] (
23:05:16linuxstbnow: I don't believe so.
23:05:54nowhow i open a port for it
23:06:03nowI can try help porting it myself.
23:07:06nowhow i open it
23:07:10nowthe devic
23:07:40nowah nevermind
23:07:44kkurbjun"Keep in mind that any posts to the public pastebin ( are considered to be placed into the public domain. So, before posting, remember that you are disclaiming any ownership of the content (it is public, after all)." Does that mean anything for us?
23:07:48 Part salty-horse ("Leaving")
23:08:49nowdam i dont have screwdriver in that size..
23:10:04 Quit sneakums (Remote closed the connection)
23:10:51 Join sneakums [0] (
23:11:33 Quit linuxstb ("ChatZilla [Firefox]")
23:12:02rasherkkurbjun: Well, in theory it does, but I'm not sure that would hold up in court
23:12:53rasherPerhaps it's time to set up a rockbox pastebin with less silly license terms "By submitting your post you grant a perpetual license to display your work on"
23:13:10rasher"you of course still retain full copyright and control over your work"
23:13:14rasheror something like that
23:13:25kkurbjunthat would work
23:14:29***Saving seen data "./dancer.seen"
23:14:30 Quit Domonoky_ (Read error: 104 (Connection reset by peer))
23:15:54 Quit miepchen^schlaf (Read error: 110 (Connection timed out))
23:15:58 Quit sneakums (Remote closed the connection)
23:16:15amiconnjhMikeS: How many yuv fps do you get on sansa?
23:17:12kkurbjunlinuxstb, how much effect do you think a hardware yuv blitter would have?
23:17:29kkurbjunoh, not here..
23:18:28 Quit BigBambi (Remote closed the connection)
23:19:06 Join SirSedgewick [0] (
23:19:29 Nick SirSedgewick is now known as Sedgewick (
23:20:30 Join BigBambi [0] (n=alex@rockbox/staff/BigBambi)
23:22:34 Join sneakums [0] (
23:23:00amiconnBuschel: Imo it would be better to put the asm helper into a separate .S file, like we did for all the other targets which have assembler optimised yuv blitting
23:24:21*scorche goes about setting up a rockbox pastebin
23:24:54 Join Mouser_X [0] (n=someone@
23:26:11pixelmadon't forget to call it "rockbin" or the like ;)
23:26:41pixelmaor "pastebox"
23:26:43bluebrotherscorche: add LaTeX hilighting to it!
23:27:02Buschelamiconn: the truth is that tried to, but i am not familiar with the .S-files
23:27:14scorchei dont have access to
23:27:42bluebrothermaybe Bagder wants to add one later?
23:28:01bluebrotherwe'll see ;-)
23:28:22 Quit now ()
23:28:47 Quit sneakums (Remote closed the connection)
23:28:56 Join sneakums [0] (
23:34:30 Join preglow [0] (i=50cb1362@gateway/web/cgi-irc/
23:34:39 Quit preglow (Client Quit)
23:34:51 Join preglow [0] (i=50cb1362@gateway/web/cgi-irc/
23:35:47 Nick preglow is now known as preglow2 (i=50cb1362@gateway/web/cgi-irc/
23:37:29 Join S1gn [0] (
23:37:44Buschelinteresting.. when you switch off the 14ms-delay in the ipod video's yuv-blit you won't see any artifacts during playback. but of course the framerates will nearly explode for low-res videos (like up to 75fps for 128x96)
23:38:19Buschel...for hi-res videos the fps is only going up little (12.5 -> 12.8 for 320x240)
23:38:38 Join makuseru [0] (
23:39:48preglow2so the yuv blit is by far the dominant cpu eater on ipod video...
23:39:48 Quit preglow2 (Client Quit)
23:39:55makuseruhi, i have a toshiba gigabeat, when i try to install the bootloader on it it just tells me its already the current version than stops the instilation
23:40:00makuseruhow can i fix this
23:40:02 Join preglow2 [0] (i=50cb1362@gateway/web/cgi-irc/
23:40:05 Quit Zagor ("Client exiting")
23:40:09preglow2argh, how i hate this web client
23:40:51S1gnhi. where to download midi instruments? and what format? can't find that in manual or rockbox website
23:41:54pixelmaS1gn: - all the information you need, I think
23:42:39bluebrotherthe wiki has a link to the patchset download so it's on the website ...
23:43:06S1gnpixelma: thanks
23:44:12*Mouser_X has some midis from about 10 years ago...
23:44:42 Quit ompaul (Client Quit)
23:45:15bluebrothermakuseru: how are you trying to install the bootloader? Using Rockbox Utility?
23:45:18Mouser_XWell, what's the file date on your current bootloader? (The one on your Gigabeat)
23:45:56makuserulemme check
23:46:28makuserubluebrother: yes
23:46:58bluebrotherwhat method are you using? Are you installing the bootloader separately or using the "Quick Start" tab?
23:47:19Buschelpreglow: could the yielding within the iPod Video's yuv-blit not give cpu-time to the mpeg-decoder?
23:47:52makuserui used the small instalation
23:48:30amiconnBuschel: You can shave off one cycle per pixel from your asm by holding the port address in a register
23:48:37bluebrotherplease try the "Installation" tab
23:48:39makuseruVersion 2.020US
23:50:02amiconnYou might need to do some rearranging in order to not run out of registers though
23:50:57preglow2buschel: only lack of yield() will result in starving the mpeg decoder
23:51:00 Quit Toxicity999 (Read error: 104 (Connection reset by peer))
23:51:09preglow2i don't really know, not too familiar with the code
23:51:37Buschelamiconn: in fact i _am_ running out of registers
23:51:55 Join Toxicity999 [0] (n=bryan@unaffiliated/Toxicity999)
23:52:00amiconnYeah, I counted 14 registers in use atm
23:52:11amiconnBut I'm sure that some rearranging is possible
23:52:19 Join preglow [0] (
23:52:40amiconnLike, you don't need to fetch Y before the chroma precalculation from Cr and Cb is finished
23:52:55 Quit preglow2 ("CGI:IRC (EOF)")
23:54:26 Quit Hukt0nWax ("I-n-v-i-s-i-o-n 2.0 Build 3515")
23:54:50 Join bryan_ [0] (
23:54:59 Quit Toxicity999 (Read error: 113 (No route to host))
23:55:11amiconnHmm, that won't help...
23:56:41Buschelamiconn: if you find a way to "find" this one register for loading the port adress please let me know :o) i also searched some minutes...
23:57:44amiconnI'm still not convinced that the general clamping check plus branch over the individual checks is faster than just running the individual checks
23:57:54*amiconn wonders whether jhMikeS tested that
23:58:10amiconnIf you drop the general clamping check, you can free a register
23:58:24makuseruthanks for the help! it works now
23:58:27Buschelbtw, deactivating the 14ms-delay in yuv-blit seems to be not the worst idea. in the worst case the current screen won't be displayed, not relevant for videos and better than being stuck while waiting.
23:58:37 Join hcs [0] (n=agashlin@rockbox/contributor/hcs)

Previous day | Next day