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

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

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

#rockbox log for 2005-09-21

00:00:23goawebguest59: use bleeding edge builds?
00:00:43webguest59Yes I use
00:00:50rasherIt's a balance act. I'd go with releasing as-is, but it's amiconn's call.
00:01:08webguest59but I think much to feature freeze :(
00:01:21webguest59amiconn, have the release button?
00:01:35webguest59the power :)
00:01:54rasherWell, right now it's the only bug being serioiusly worked on. I think the rest are going to be left.
00:02:25Danh120 arrives tomorrow
00:02:43webguest59but why don't release if all is good
00:03:00rasherBecause it'd be nice to get this bug fixed first.
00:03:12webguest59yes indeed
00:03:37webguest59it will be fixed in the daily builds
00:04:27webguest59and tell to user that there is 1 bug left, please wait the bug will be fixed in the daily builds or still one release
00:04:47webguest59*new one
00:05:26rasherYes. That is one solution.
00:05:30rasherBetter would be to get it fixed.
00:06:00webguest59yes but the time run fast
00:06:10linuxstb_webguest59: Everyone is still hacking away at Rockbox - there will be a _lot_ of changes committed to Rockbox when the feature freeze is over. Development hasn't stopped.
00:06:36webguest59in the rockbox. org Bagder's said release around 5 septembre :)
00:06:52webguest59linuxstb: ok
00:07:12Bagderwebguest59: that was the plan
00:07:39webguest59plan have changed?
00:07:55webguest59you can modify this message no
00:08:02Bagderthe plan was to "attempt" to release it at sep 5
00:08:08 Quit Dma-Sc ("")
00:08:22webguest59optimistic way?
00:09:14Bagdernot really
00:09:27Bagderjust not many people involved in the bug fixing process
00:09:37Bagderand this recording bug being a lot meaner than expected
00:09:38rasherI'm curious if amiconn has any other ideas lined up. If not, I'm of the opinion that the release should go ahead.
00:09:46Bagderrasher: I am too
00:09:52webguest59have you got one limit time, exemple, if the bug not fixed the X sept, october...?
00:10:16rasherwebguest59: No. Time limits don't work well for stuff done on people's free time.
00:11:18rasher(as has been demostrated brilliantly)
00:11:20webguest59yes indeed, I thought about administration way of rockox
00:12:00webguest59I thanks everyone here again for all the free works done
00:12:14webguest59I'm not complaining, just wondering
00:12:27***Saving seen data "./dancer.seen"
00:15:40webguest59rasher: for the WPS gallery wiki section, maybe you could add message in forums gallerys for the owner will can add them again to the wiki
00:16:22rasherI guess I should
00:16:33webguest59very good effort you did everyone can thank you for this
00:16:59webguest59assume a lot of hours at restored :(
00:18:14Danyou lot are gunna love helping me get rockbox setup tomorrow
00:18:18Dancant wait
00:18:25 Quit Dan ()
00:22:26 Join paugh [0] (n=kickback@2001:5c0:8fff:ffff:8000:0:3e03:6822)
00:24:07webguest59bye all
00:24:25webguest59and I hope realese coming soon ;)
00:24:30 Quit webguest59 ("CGI:IRC")
00:37:58 Quit ender` (Read error: 113 (No route to host))
00:39:11rasherHaha, good one Febs. (on the mailing list)
00:43:36 Join thegeek_ [0] (
00:47:26 Quit thegeek (Read error: 113 (No route to host))
00:47:56 Join thegeek [0] (
00:48:43 Join ashridah [0] (
00:49:34 Quit Moos ("Parti")
01:02:48preglowabout time as well
01:03:31rasherLong due
01:05:46 Quit thegeek_ (Read error: 110 (Connection timed out))
01:06:12rasherInteresting stuff, this Neuros talk
01:06:50Bagderpreglow: in #haxx
01:06:51rasher#haxx, you missed by a few hours though.. still going on though
01:07:09preglowoh, right
01:07:13preglowdidn't know it was today
01:07:16rasherThere's preglow long-awaited ARM platform!
01:07:19Bagderlog will be available
01:07:38linuxstb_rasher: No, preglow's going to buy an ipod.
01:08:03rasherYeah, mostly joking
01:08:15fuzziemmm, rockbox on ipod.
01:08:23linuxstb_rasher: Me too.
01:09:04preglow_200_ mhz arm9?
01:09:10preglowthen what the flaming hell do they need the dsp for
01:10:05*preglow starts on the ti toolset woes
01:10:27Zagorbut basically the idea is to not use the DSP on the audio players
01:10:54Bagderthey want the same CPU for several models
01:11:01preglowahh, i see
01:11:05preglowoh well
01:11:15preglowadapting existing code for dsps is a nightmare anyway
01:11:32 Join edx [0] (
01:15:19DeepBso... i've heard you had problems with TWiki, right?
01:16:15rasherWe did, yeah uses TWiki also, anything i should know?
01:16:43fuzziemake sure you applied the early-september security patch
01:16:50fuzziewhich is mysteriously not mentioned anywhere on their front page
01:17:34rasherthat one
01:17:57Bagderand check your server for alien processes
01:18:33 Join elinenbe [0] (
01:18:34Bagdergrep apache logs for "?rev=.*% HTTP"
01:18:46Bagder"?rev=.*%.* HTTP" even
01:18:56elinenbewhat sort of cunt destroys a free open source project.
01:19:08elinenbesorry for the language...
01:19:12elinenbebut what an ass.
01:19:28fuzzieprobably someone randomly scanning for twiki installs via google or something, to be honest
01:20:09rasherThat's my guessing too
01:20:11Zagorfuzzie: yes
01:20:28rasherI fail to understand the reasoning behind this
01:20:35preglowas for dsp chips and gcc support: are there any?
01:20:42elinenbehas Linus worked anymore on the H320?
01:20:42preglowi know blackfin's god gcc support
01:20:51Zagorpreglow: not really
01:20:52preglowbut blackfin isn't exactly a typical dsp
01:21:07preglowand it's 16 bit
01:21:32preglowto get the most out of your typical dsp chip, you need to use asm
01:22:33DeepBwell.. it's hosted on
01:22:40 Join matsl [0] (
01:22:46BagderDeepB: then you're probably safe
01:22:52Bagderyou can easily check it
01:23:02DeepBso i don't have to worry about the server
01:23:04Bagderthere's a sample URL in that alert
01:23:10DeepBbut i appreciatte my data
01:23:31Zagorthe sample url is harmless
01:23:37fuzziejust make sure you're patched
01:23:40DeepBthank you for the pointers guys, i'll check it right on
01:23:48fuzziebut, to be honest, you should be backing up stuff from sourceforge regularly
01:24:10fuzziethere's nothing to stop other sf users from damaging anything apache-writable, afaik.
01:24:33Bagderunless they run suexec
01:25:13fuzziethey didn't when i was last hosting web stuff there
01:25:23fuzzie[i've long since deserted them, cvs is unusable]
01:25:23 Quit matsl (Remote closed the connection)
01:25:37 Quit [IDC]Dragon ()
01:25:44Bagderyes, I've abandonded them too for almost everything
01:25:47*Zagor has painful memories from when rockbox had cvs at sf.
01:26:24rasherNow someone write that SF > Bojira migration script
01:26:50rasherSo we can ditch the patch/bug tracker
01:26:55rasherIt's vile.
01:27:39ZagorI agree very much.
01:27:47Zagorbut I've never heard of Bojira :-)
01:27:51 Quit ashridah ("Leaving")
01:27:57fuzziebojira being bugzilla :)
01:28:00rasherOh, Bugzilla
01:28:13DeepBfuzzie: i know, i'd like to get rid of that can of worms as soon as i can
01:28:25fuzziehm, doesn't point at the right pages any more
01:28:44rasherfuzzie: crikey
01:29:09fuzzieexcellent :)
01:29:22ZagorI'm having problems registering on their mailing list (and another list)
01:29:34BagderI managed just now
01:29:49rasherMaybe SF is just slow.
01:29:55rasherWho am I kidding, SF *is* slow.
01:29:56fuzzieoh, right, they can turn off the power on the DSP, that's useful
01:30:34 Quit linuxstb_ ("Leaving")
01:30:36Zagorrasher: slow as in four days?
01:31:06fuzziesure your mail server isn't blocking them?
01:31:18fuzziei signed up to the neuros442linux-main thing and it only took a few minutes
01:31:18 Quit paugh (Excess Flood)
01:31:37Zagoryes. I get the confirmation mail, but when I confirm I don't get the welcome mail.
01:31:54 Join paugh [0] (n=kickback@2001:5c0:8fff:ffff:8000:0:3e03:6822)
01:31:55BagderI bet they enabled the Zagor filter
01:31:59BagderI use it all the time ;-)
01:32:03Zagorthey too? gaah
01:32:35Bagdernow, less than 5 hours to get-up time, I better sleeeeep
01:32:45fuzziewell, i've managed to get both, total time between them 3 minutes or so, so i guess they might well have done :)
01:33:06ZagorI'm feeling discriminated. ItProbably
01:33:07rasherI haven't got my confirmation mail yet.
01:33:24ZagorIt's probably because I'm.... uh... tired.
01:33:34rasherBut then, I think some server along the way to me is doing greylisting and not being very smart about it
01:34:14*preglow ponders the possibilities of having a proper dsp core...
01:34:58Zagorpreglow: the possibilities are great, if you have some way of programming it. doing it all in assembler is not too fun...
01:35:24fuzziedepending on your idea of fun
01:35:24preglowi have somehow gotten used to assembler since starting to code for rockbox...
01:35:30 Join void [0] (
01:35:33Zagorpreglow: hehe
01:35:38voidheya guys
01:35:48voidwas wondering if there's a site or area I can download wps files from?
01:36:03voidit's for the archos recorder
01:36:06preglowvoid: you're out of luck, our wps page is broken after being hacked
01:36:20rasherAh, for the recorder it should be up to date
01:36:26preglowgoodness, then
01:36:27Zagorpreglow: the WPS:es are in the text, not as attachments">
01:37:57rasherThat's linked at the top of the page anyway
01:38:01voidcool thanks a ton :)
01:44:09 Join linuxstb_ [0] (
01:44:53 Quit hicks (Remote closed the connection)
01:55:39 Quit paugh ("Leaving")
01:55:44 Join paugh [0] (n=kickback@2001:5c0:8fff:ffff:8000:0:3e03:6822)
02:07:30 Part stripwax
02:10:21 Quit preglow ("leaving")
02:12:32***Saving seen data "./dancer.seen"
02:48:33 Quit tvelocity ("Leaving")
02:51:41 Join lee_ [0] (
03:09:03 Quit ze (Read error: 104 (Connection reset by peer))
03:13:41 Join bitshift [0] (
03:21:05 Join dropandho [0] (
03:21:12 Part bitshift
03:21:16dropandhorasher- u rock
03:21:20dropandhothanks for all your work
03:21:26dropandhodo you need any help?
03:21:34rasherI think it's mostly done
03:21:45dropandhoand how much do you get paid for this?!
03:21:53rasherThe WPS gallery is left - leaving that for people to fix by themself
03:22:04dropandhoyeah- that is a good call
03:22:07rasherdropandho: I've been paid one pice of excellent firmware
03:22:20dropandhodo you wanna announce your progress on the front page
03:22:25dropandhohehe- i hear ya
03:22:50rasherHm, maybe there should be something
03:23:20dropandhojust so folks know whats up
03:23:30dropandhoespecially the wps people
03:23:48rasherI've marked the WPS page quite heavily
03:23:59dropandhogot it
03:24:26dropandhomaybe just about the success
03:24:32dropandhowhat is missing and what isnt?
03:24:40dropandhoplus, you need to get props
03:25:17rasherHah, I'll do fine without.. but let me add something..
03:25:37 Quit paugh ("Leaving")
03:25:39rasherNow I know the wiki quite well after this..
03:26:20dropandhono kiddin
03:26:47dropandhoand also maybe something about if the hole was patched up
03:26:52*rasher checks out www
03:26:55rasherOh, it was
03:27:03dropandhoand the progress on fixing a regular backups
03:27:06rasherBefore going online again
03:27:11 Part DeepB
03:27:15dropandhoi dont know...just thinking about what inquiring minds might wanna know
03:27:45rasherFair enough.. I'm going to add a note
03:27:49rasherin.. uh.. a while, it seems
03:28:53dropandhofair enough
03:28:57dropandhoso no help is needed?
03:29:20rasherNot outside the WPS gallery really
03:29:28 Quit lee_ ("Leaving")
03:29:35 Join bagawk [0] (n=lee@unaffiliated/bagawk)
03:29:40dropandhoimpressive...thanks again
03:30:43rasherWell, I didn't do it all alone. Probably did most of it though I guess
03:31:26dropandhoyes, get some sleep now!
03:32:06dropandhook- have a good night
03:32:13dropandhothanks and take care rasher
03:32:29rashernight, thanks for the appreciation
03:32:31 Quit dropandho ()
03:37:51 Join ze [0] (
03:37:55 Quit ze (Read error: 104 (Connection reset by peer))
03:42:30 Join ze [0] (
03:42:37 Quit ze (Read error: 104 (Connection reset by peer))
03:43:31 Join ze [0] (
03:43:36 Quit ze (Read error: 104 (Connection reset by peer))
03:44:41 Join ze [0] (
03:44:43 Quit ze (Read error: 104 (Connection reset by peer))
03:45:41 Join ze [0] (
03:45:44 Quit ze (Read error: 104 (Connection reset by peer))
03:53:15 Join ze [0] (
04:06:08 Join QT_ [0] (i=as@madwifi/users/area51)
04:12:36***Saving seen data "./dancer.seen"
04:13:31 Quit bagawk ("Leaving")
04:16:00 Quit QT (Read error: 110 (Connection timed out))
04:28:07 Join Mirthoneist [0] (
04:28:07 Quit Mirthoneist (Client Quit)
04:30:41 Join Mirthoneist [0] (
04:50:47 Quit cYmen ("zZz")
05:05:40 Join Yono [0] (
05:07:19 Quit Yono ("Download Gaim:")
05:13:39 Quit rasher (Remote closed the connection)
05:48:57 Quit Mirthoneist ()
06:00:46 Join Tomas_ [0] (
06:12:39***Saving seen data "./dancer.seen"
06:15:33 Quit t0mas (Read error: 110 (Connection timed out))
06:15:33 Nick Tomas_ is now known as t0mas (
06:20:14 Join cheriff [0] (
06:20:27 Quit edx ()
06:33:33 Join LinusN [0] (
06:39:27amiconnLinusN: Did you check ? Your work done for soft poweroff and panic is missing; I just changed the xing header status
06:39:59LinusNdidn't look, i'll fix
06:43:39LinusNi was thinking about the bit shift
06:44:21LinusNare there occasions where we disable the interrupt for a long period? i think not
06:44:32amiconnNone that I know of
06:45:01amiconnThis bitshifting has some really puzzling properties
06:48:12 Quit Mxm`Pas`Bien (Read error: 104 (Connection reset by peer))
06:48:44 Join amiconn_ [0] (
06:48:55 Join Maxime [0] (
06:52:00 Join amiconn__ [0] (
06:53:32LinusNamiconn_: puzzling properties?
06:57:24 Quit amiconn (Nick collision from services.)
06:57:24 Nick amiconn__ is now known as amiconn (
06:57:29 Quit amiconn_ (Nick collision from services.)
06:58:08amiconnLinusN: Yes. First, the probability of a bitshift happening depends on the transfer speed used.
06:58:33LinusNthat suggests performance problems
06:58:52amiconnFor the cvs C loop (and my local, slightly more optimised C loop) it happens after 2..4 hours
06:59:28amiconnWith my highly optimised asm loop (better than what was in 2.3 and 2.4) it happens after 30 min .. 1 hour
07:00:07amiconnThis is strange, since it would suggest that it would happen really often when the MAS is connected to a real DMA controller
07:01:13amiconnThese time periods are rather long compared to the amount of data transferred - one shift after millions of bytes
07:01:43amiconnSecond, the shift seems to always happen at a frame boundary, which is not necessarily a DMA block boundary
07:02:09amiconn(although I'm not 100% sure about the frame boundary)
07:03:03LinusNno, that's not easy to tell
07:03:39LinusNso maybe it is a reverse performance problem
07:03:52LinusNthat the mas can't cope with out fast transfer speed
07:06:26amiconnOh, and third, the probability depends on the quality setting and the recording level
07:07:14LinusNalso suggesting a mas performance problem
07:07:46amiconnI derived the frame boundary rule from my simple restore method. As it's not easy to tell the exact point, I presumed it was at the frame boundary, and undid the bitshift accordingly with my tool
07:08:18amiconnWhen playing the recording, there were no audible glitches at these points
07:09:06amiconnI suspect archos might know some more about the mas recording init than what the datasheet says
07:10:40amiconnI don't think it's a simple performance problem. (1) The mas tells us when it's ready to transfer a block, by raising EOD. Why would it do that before it's really ready for transfer?
07:11:21amiconn(2) The bitshift isn't temporary for one block or one frame, it stays for hours until another shift happens eventually
07:12:01amiconnI even had one test recording sequence where the next shift brought the bit order back to normal....
07:13:00amiconnIt seems that the mas tends to shift forward, the most observed shift values are +1 ... +3 bits (i.e. to the right)
07:13:30amiconnAlthough I had one case of -1, I believe this really was a +7 shift
07:14:35amiconn...derived from the fact that I also had one case of a +8 shift, which of course didn't destroy the recording, but there was a stray byte in the stream...
07:18:02LinusNso perhaps the internal transfer from the DSP to the dma buffer inside the MAS is serial
07:18:43Bgermorning all :)
07:24:50Bgeri see rasher has completed the wiki update ...
07:28:47amiconnLinusN: I don't know whether it's only a rumour, but didn't archos make the mas recording at 192 kbps cbr is some later product?
07:29:07LinusNhaven't heard of it
07:29:40amiconn192 kbps cbr means more load than q=7 (which goes up to 192 kbps frames, but is ~170 kbps on average)
07:31:15LinusNhmmm, i see we poll EOD
07:31:45amiconn around 12:30
07:32:08LinusNaccording to the data sheet, EOD might not go inactive until after 8us
07:33:37amiconnYes, and that's what the timeout loop for /rtw is for
07:34:04amiconnAs EOD might still be high when the DMA buffer is empty, we raise PR as we expect another byte
07:34:14amiconnA real DMA controller would do the same
07:34:59LinusN"New Feature: MP3 encoding can now be done in 192kbit/s CBR for sample rates of 32, 44.1 and 48kHz"
07:35:03amiconnThen we wait for /rtw to signal the next byte is ready. If it's not (after a much longer timeout than 8Ás) we lower PR again
07:35:42amiconnThe archos recording loop does the same, with an even longer timeout
07:35:53amiconn(which doesn't make a difference; I tried)
07:37:08 Join B4gder [0] (
07:39:32LinusNi suspect a real dma controller would wait for _RTW before polling EOD
07:40:44B4gderseen the neuros chat log?
07:40:50LinusNB4gder: no
07:41:06LinusN(i fell asleep ans missed the meeting, sorry)
07:41:49amiconnLinusN: Nothing guarantees that EOD goes low before the final /rtw -> inactive transition (according to the datasheet)
07:43:50amiconnNo. If we lower PR after t3, EOD might still be high because teod may be up to 8 Ás
07:44:06amiconnPerhaps a rather unusual approach would work - keeping PR high by default (!)
07:47:45LinusNyes, if we lower PR after t3, EOD might still be high, but i doubt it would still be high when RTW goes up again
07:49:47amiconnIirc the very old loops had that (in rockbox 2.2), but I'll do a test
07:53:23amiconnTest running, gotta hurry now...
07:57:33 Join Febs [0] (
08:01:16 Join ender` [0] (i=ychat@
08:12:43***Saving seen data "./dancer.seen"
08:14:58t0mas(and LinusN and Zagor)
08:15:03t0masreading the logs about neuros...
08:15:13LinusNme too
08:15:20t0masmaybe it's a good idea to register rockbox as a company
08:15:34t0masor isn't it possible to register a non-profit organisation in your country?
08:15:40B4gderyes it is
08:15:48t0masmight be a good idea to do that?
08:16:07B4gderperhaps, I don't think it'll change much
08:16:15t0massetup a contract where you three are the owners and delegate the copyright to the writers
08:16:29t0masbut add a part explaining that you are the last right holders
08:16:45B4gderwell, it doesn't work in quite that way
08:17:04t0masit does if you ask everybody with cvs write access to agree to it
08:17:30B4gderyes, but they would have to agree to sign over their copyright to us
08:17:47B4gderand I don't think we need that
08:17:49t0masin holland it can be done different...
08:17:56B4gderI doubt that
08:18:20B4gdereach author owns their own work
08:18:28t0maswell... we have some system here... making it possible for all authors to have the rights, but at the same time giving 1 company the right to sell it etc...
08:19:16t0masand if they do... the authors still have the copyrights on the version at that point... but the new version is free for the next company to do whatever they want with it afaik
08:19:41B4gderin this case, no one can forbid any company to sell rockbox
08:19:56t0mas(don't think it's GPL compatible... it's used in universities a lot)
08:20:31B4gderI don't see how we have a problem, so I don't see anything to fix ;-)
08:20:47t0maswell... you can't contact Ti now for example...
08:21:04t0masas said in the logs?
08:21:08B4gderno one can get any sense out of TI
08:21:21B4gderNeuros been trying for years
08:21:36t0masand for hosting contracts etc... it's 1 of you 3 who's the contractee...
08:21:56t0masand afaik you have to pay taxes on the gifts to rockbox?
08:22:03t0masit is that free in your country?
08:22:33B4gderit is free
08:22:44B4gderdonations are
08:23:18t0masah ok
08:23:29t0masthat makes it a lot easier
08:23:56t0maswell.. maybe it's different there then... in holland we have some regulations madness...
08:24:03t0masso you have to register almost anything
08:38:46Febst0mas and JoeBorn raise some interesting points about the legal status of Rockbox.
08:39:45B4gderI don't think it is primarily a legal problem
08:41:32B4gderother than perhaps a question about who owns the name and the logo etc
08:41:39FebsWell, for example, who has authority to sign a contract on behalf of Rockbox, should a contract need to be signed?
08:41:49B4gderhm, no not the logo
08:42:07FebsActually, the logo is a good example. Who designed the logo?
08:42:16B4gderthe logo is owned by its creator
08:42:22Zagorit's pretty much the same using rockbox or linux. who signs contracts for linux? nobody does.
08:42:31t0masFebs: nobody can sign anything on behalf of rockbox now
08:42:37t0masbecause rockbox legally doesn't exist
08:42:46B4gderThat isn't necessarily true
08:42:54Zagorrockbox does exist. it's just not a company.
08:42:58t0masZagor: Linus (torvalds) started some holding company for linux...
08:43:01B4gderan informal team exists
08:43:06B4gdereven in the eye of the law
08:43:22t0masit doesn't here...
08:43:37*B4gder is involved in talks about exactly this in a very high-stake open source project
08:43:43t0masyou can register some legal entity for informal non-profit things like sportsteams etc
08:44:13t0masintresting how this is done different all over europe...
08:44:29Febs... and here on the other side of the pond!
08:45:10Zagorthe point is the law isn't blind. nobody is allowed to railroad over us just because we're not a company. individuals have the same rights as companies.
08:45:39t0masyes, we have... but we can't do anything (like getting a hosting contract for the website) as rockbox...
08:45:51t0masthat has to be done as some individual... who is responsible for it then
08:45:51B4gderI'd say that the biggest problem would be our international nature
08:46:06t0masyeah, that is a problem I guess
08:46:19B4gderwe'll simply don't sign contracts as "rockbox"
08:46:46B4gderat least, not until we have a formal org for it
08:46:47Zagort0mas: I don't see that as a problem (contracts). in the end single individuals *are* responsible.
08:46:55Zagorwe wouldn't be better off having "rockbox" responsible
08:46:56t0maswait a moment
08:47:03t0masclient fucking up again
08:47:12FebsZagor, yikes!
08:47:16t0masah, back
08:47:35t0masZagor: got to run
08:47:47FebsDo you want to be individually responsible if, for example, Rockbox bricks a large number of players?
08:47:58ZagorFebs: yes
08:48:07B4gderyes, if you sign a personal contract that you are, then sure
08:48:20t0mas(sorry... have to work now)
08:48:45Zagorhowever we put pretty strong disclaimers on the firmware that says "works for us, might not for you".
08:50:56 Join aga [0] (
08:51:47agahey does rockbox 2.4 supposed to display a 'charging' screen when you have the charger plugged in
08:52:42agai just got a used archos and am trying to determine if its got a fault
08:54:10Zagoraga: which model?
08:54:22agajb recorder 20
08:54:45agathe battery icon will have a question mark sometimes, then after a while it says 40% or so
08:55:04Zagorwith the charger plugged, the battery icon should at least animate
08:55:14agaand ive changed the batteries to ones that i know are good
08:55:55agaok so i have either a bad unit or a bug i guess
08:56:32Zagorcould be a broken dc/dc regulator. that's not entirely uncommon.
08:57:09agadoes that workaround with the F1 have anything to do with this?
08:57:36agaoh you mean the charger might be bad, or the connecting pin
08:58:11Zagoran IC in the archos. it breaks if you insert a charger with wrong polarity or such.
08:59:26*Febs has been thinking about Zagor's point and disclaimers and B4dger's point about personal contracts.
08:59:37agai saw that faq in the forum, i guess you get the IC from archos
09:00:13agaanyway to quickly tell if that is the problem besides disassembling
09:00:28Zagoraga: it's a pretty standard part you get from an electronics shop.
09:00:30FebsThe whole nature of an open source project raises some really interesting legal questions.
09:00:43FebsPerhaps more interesting to me than to you folks.
09:01:09BgerFebs are you a lawyer ? ;)
09:01:16 Join einhirn [0] (
09:01:19B4gderI am quite interested in the legal parts as well
09:01:54agai didnt understand that 'press F1 ON' sequence then plug in the charger bit. does that relate to my problem?
09:02:25Zagoraga: no, that just makes you boot the archos firmware instead of rockbox
09:03:12agaok thanks, ill check the faqs and see whats up
09:04:05ZagorFebs: start at the other end. since when are commercial software companies responsible for their bugs? microsoft software bugs causes billions of damage every year, for instance.
09:05:26 Part aga
09:06:37 Part cheriff
09:07:58BgerZagor have you read the M$'s EULA ?
09:08:12ZagorI have
09:08:28Zagorwell, a few of them. different products use difference eulas
09:09:11Bgerfrom "diagonal" reading, afaics, they wash away their hands ...
09:09:27Bgerabout responsibility
09:09:43Zagoroh yes, every way they can (and some they can't)
09:10:49 Join aga [0] (
09:11:01agaI forgot one thing
09:11:32agawhen you boot up, are you supposed to see the firmware 1.8, *Then the rockbox logo afterwards?
09:11:41agaon the jb recorder 20
09:11:52Zagoraga: if you are not flashed, yes
09:12:15agaoh ok, so that is normal, not a conflict between programs
09:12:34Zagorno. the archos firmware boots and then finds rockbox and loads it.
09:13:12agadang, i thought that would be it
09:13:25*aga slogs off
09:13:28 Quit aga (Client Quit)
09:14:32FebsBger, yes. I am a lawyer. (A lawyer with insomnia at the moment.)
09:14:46Bgerhehe :)
09:14:53Bgerthat's not good
09:16:25FebsYeah, it's 3:14 AM here in the eastern U.S.
09:17:41FebsI'm also a working musician.
09:17:57Bgerwow:) interesting combination
09:18:19FebsYeah. Not always easy to balance.
09:18:52ZagorBger: the GPL (and pretty much every other license) contains such No Warranty clauses too, however not quite as far-fetched as the MS EULAs.
09:19:26Bgerand at least in most cases you pay for support, not for the product itself ...
09:20:14Zagorthat's an issue too: even if authors were liable, how much warranty does your $0 buy you?
09:21:09Zagorrequiring overreaching liability from authors would completely kill free software
09:22:01ZagorCan you tell I too am interested in the legal angles? ;)
09:22:32FebsLike I said before, many interesting questions ...
09:23:51Febs... which I've been very careful not to answer. ;)
09:24:04Zagorhehe. you ARE a lawyer! :-D
09:28:32FebsIn all seriousness, as I'm sure you can appreciate, I *do* need to be careful not to give legal advice in a public forum.
09:31:43ZagorLet's just say if free software liability would become an issue, Rockbox is pretty far down the list of prominent projects getting in trouble.
09:32:18Zagor(Heck, half the Internet would probably shut down... )
09:33:38 Join bobTHC [0] (n=bobTHC@
09:33:44bobTHCmornin' folks !
09:34:25bobTHCi'm fine and u ?
09:34:31FebsIt's good night for me. Time for some sleep ...
09:34:41ZagorFebs: good night
09:34:46ZagorbobTHC: just peachy
09:35:03bobTHCnice ;)
09:35:18 Quit Febs (" HydraIRC -> <- Go on, try it!")
09:43:16 Join Dma-Sc [0] (
10:02:39 Quit linuxstb_ ("Leaving")
10:12:47***Saving seen data "./dancer.seen"
10:26:37 Join DangerousDan [0] (
10:37:27 Join ashridah [0] (
10:43:08Bgerjoke of the day: haha
10:43:17Bgerdon't try it in IE
10:44:57Bgerbtw, i gave up translating rockbox to bulgarian. The system font doesn't have cyrillic (cp1251) chars, so i can't do it as it should be
10:45:18snaxapparently my system language is "undefined"
10:45:31Bgersnax and your processor ? :)
10:45:34snaxwhich I speak quite fluently, I might add
10:46:16B4gderyour processor family is "undefined"
10:46:41 Quit DangerousDan ("Miranda IM! Smaller, Faster, Easier.")
10:46:42Bgerand it uses ActiveX for displaying explorer window
10:46:47snaxapparently I also live in Bellevue, United States
10:47:08snaxI must have missed it when that city (which I don't live in, by the way) became a state in the union
10:48:40snaxwords such as "boy" "father" and "child" were found in my system records
11:09:09 Join zeekoe [0] (
11:10:56 Join tvelocity [0] (
11:15:41 Join preglow [0] (
11:28:55 Join Moos [0] (
11:38:55HClcan anyone tell me whether rockbox will play audio from an mpeg2 file?
11:39:56preglowmpeg2 allows for aac audio as well
11:40:02preglowand that we can't play
11:40:09preglowbut we can play mpeg2 layer 3 streams
11:40:25HClbut if i have an mpeg2 video file
11:40:34HClwill it take the audio stream?
11:40:45preglowdepends on what it is
11:40:53HClnot quite sure, let me check
11:48:03 Join cYmen [0] (
11:52:36*HCl taps his foot while mplayer compiles
12:10:10preglowlong-whinded, is it?
12:10:57 Join Vladoman [0] (
12:10:59Bgerwas it on dual xeon 500 or sth similar ?
12:11:28Vladomanamiconn, you still there?
12:11:42Bgerhe is offline atm
12:12:15Vladomananybody here working on the "bit shift" problem
12:12:49***Saving seen data "./dancer.seen"
12:13:11preglowonly hiom
12:14:19Vladomancause I can give some info on it
12:14:34Bgerjust shoot it
12:14:42Mooslog ;)
12:14:45Bgerthere are irc logs, and he'll see it
12:15:38Bgerbtw, see
12:17:52 Quit zeekoe (" HydraIRC -> <- Try something fresh")
12:17:55Vladomanyep, I just mailed him anyway
12:41:12 Nick QT_ is now known as QT (i=as@madwifi/users/area51)
12:48:47 Join am1conn [0] (
12:49:29Bgeram1conn how was the test
12:50:37am1connStill running @home, can't check atm
12:51:42MoosVladoman: still here?
12:59:18 Quit Dma-Sc ("Wow! What a great client! Bersirc 2.2 [ - Open Source IRC ]")
13:11:28 Join linuxstb_ [0] (
13:11:56linuxstb_HCl: If you use mplayer to extract the audio from an MPEG2 file (I'm guessing it's a "program stream"), then Rockbox should play it.
13:12:11linuxstb_It's most likely either MP2 or AC3 audio. Use "mplayer -dumpaudio"
13:13:49preglowyeah, can't say i've ever seen aac in an mpeg2 stream
13:14:29preglowany chance of seeing dts support in rockbox? ;)
13:14:41LinusNwhat is dts?
13:14:54linuxstb_I read on misticriver that someone played DTS via the optical out on the H1x0.
13:15:07linuxstb_LinusN: It's a compressed multi-channel format.
13:15:12linuxstb_Similar to AC3
13:15:54linuxstb_"DTS in WAV" is a common format - the player treats it as a normal WAV file, and passes the bitstream to an external decoder via S/PDIF
13:16:07preglowto say it's similar to ac3 isn't exactly true
13:16:09preglownot internally, at least
13:17:07linuxstb_I just meant that it's a lossy compression format designed for multi-channel audio. Is that not true?
13:17:27preglowoh, yes
13:18:10preglowbut they're pretty different codecs, i don't think dts is a transform codec at all
13:18:25preglowwhereas ac3 is more or less just an mdct and some magic
13:18:44preglowi think i've seen some libdts some place
13:19:59LinusN"mdct and some magic", pretty much covers it all :-)
13:20:24LinusNyou really make it sound simple ;-)
13:21:18linuxstb_I wouldn't know - I just treat codecs as black boxes.
13:22:07preglowwhat, 'magic' isn't a professional term? ;)
13:23:42preglowbut no, dts is an adpcm like codec, i think
13:24:07 Join DangerousDan [0] (
13:31:52VladomanMoos: Yes
13:32:11BgerVladoman am1conn is (was) here
13:33:21Vladomanthe MAS bit shift is a comfirmed bug (by Micronas), there is no "fix" that they know of (AFAIK)
13:33:51preglowthen how come the archos firmware can record for quite a while with no shift?
13:34:19Vladomanmaybe by working around it?
13:35:08preglowa lot of factors play in, apparently
13:35:15preglowlike transfer speed
13:37:07Bgermaybe they're scanning the stream from mas realtime while recording (sorry if i'm talking bullsh*ts))
13:39:32preglowwell, they are
13:39:48preglowbut i don't know how that helps them in detecting a shift
13:40:07preglowperhaps it is possible to detect a shift in the sync bits
13:40:24Bgeri wonder that also
13:40:28preglowi wonder if the bit shift can happen anywhere in a frame or at a boundary
13:40:55Bgeramiconn said that he observed the bitshift *only* at frame boundaries
13:41:01Bgersee the irc logs
13:42:03Vladomanif the bits are shifted, the sync words would not match any more, no?
13:42:38preglowindeed not
13:42:50Vladomanso it can be detected
13:43:00 Join tuco [0] (
13:43:03preglowunless the bit shift is eight bits
13:43:17preglowin which case the only thing you can see is a skipped byte
13:43:23tucoIs this the same bit shift as amiconn is investigating?
13:44:35Vladomanyes, but a skipped byte can also be detected
13:44:54preglowyes, sure
13:45:09preglowthe workaround they describe here isn't very nice, though
13:45:13tucoand the workaround described doesn't work?
13:45:15preglowwill probably mean some lost recording time
13:45:27tucook, probably :)
13:45:31preglowamiconn has considered this workaround
13:45:38preglowdon't know about the parity checking, though
13:46:00tucoI suppose he have.
13:47:17Vladomantuoc: this SPDIF issue is known, it's not the same thing
13:47:27tucoVladoman: ok
13:47:51Bgerpreglow today
13:48:01Bger's log, 07.01.43
13:48:40 Join markun [0] (
13:50:18tucoHave to say amiconn goes through this issue thoroghly
13:51:11preglowyes he does
13:51:25tucowhich is good
13:51:38 Part crash3m ("Leaving")
13:53:11 Part tuco
13:53:33 Quit tvelocity ("Leaving")
13:57:06 Quit Bger ("BitchX-1.1-final -- just do it.")
14:09:32 Join Bger [0] (n=Bager@
14:12:46 Join [IDC]Dragon [0] (
14:12:52***Saving seen data "./dancer.seen"
14:12:59[IDC]DragonHi Vladoman ;-)
14:13:14Vladomanhi mighty Dragon
14:26:19 Part Vladoman
14:28:41 Quit am1conn ("CGI:IRC (EOF)")
14:31:52 Quit B4gder ("time to say moo")
14:33:13Bgerhehe good quit msg :)
14:36:35 Part LinusN
14:43:26 Join LinusN [0] (
14:45:23 Part LinusN
14:45:52 Join am1conn [0] (
14:46:32[IDC]Dragonyou got some info? ;-)
14:47:01am1connYes I did
14:47:28am1connHowever, there's no quick way to implement that in rockbox
14:48:07[IDC]Dragonyes, we need Luke Framewalker
14:48:13 Join LinusN [0] (
14:48:32preglowam1conn: have you read what i said about detecting bitshift in a frame? an hour ago or something
14:48:47[IDC]Dragoncould do some good on playback, too (audible FF/FR)
14:49:29[IDC]DragonLinusN in Germany \o/
14:49:47LinusNleaving tomorrow
14:52:04 Join tvelocity [0] (
14:54:06 Join rasher [0] (
15:01:54Bgerrasher congrats for the work yesterday
15:02:44rasherWhy, thanks.. there were not that many pages with big changes left
15:03:34 Join crash3m [0] (i=crash3m@unaffiliated/crash3m)
15:03:46Bgerhm, at least IriverInstall :)
15:05:42rasherYeah, that took a while
15:06:24preglowtrivial commit time
15:06:34 Part crash3m ("Leaving")
15:07:10Bgerhm, it seems that the new wiki doesn't allow to create attachments without comment
15:07:23LinusNyes, we're aware of that
15:07:26LinusNsilly indeed
15:07:47LinusNand the "view raw" is horrible
15:08:24Bgerit's horror to read bigger pages
15:08:51rasherShouldn't take much to fix
15:09:43Zagori'll look at both issues in a bit
15:10:58LinusNZagor: it seems you'll have to change the wiki source code
15:11:11LinusNyou did that the last time too?
15:11:32Zagoryes, but I'd like to avoid it this time since we're using auto-updated packages
15:11:39Zagorwe'll see
15:11:58LinusNit doesn't seem to be configurable, from what i saw in the code
15:16:47 Join atubbs [0] (
16:05:42 Quit am1conn ("CGI:IRC (EOF)")
16:10:30 Join Rori [0] (
16:10:35Rorihi. anyone awake?
16:10:56 Quit Moos ("Parti")
16:11:07rasherCould be
16:11:26RoriI have a couple of problems with latest builds of h120 rb
16:11:39LinusNno, you're lying
16:11:49LinusNthere is noooo problem in rockbox :-)
16:12:03*LinusN looks away
16:12:52Rori1. I get a dir buffer full error when accessing a directory with over 1000 files in it. 2. I can't get wps to display correctly anymore. It shows the bmp path and fonts don't change. Also it seems to crash on occasion
16:12:54***Saving seen data "./dancer.seen"
16:13:08BgerRori increase dir buffer size
16:13:11RoriI did
16:13:14Rorimakes no difference
16:13:17LinusN1) Adjust the "Max files in dir" setting
16:13:25RoriI set dir limits to 10,000 same prob
16:13:28LinusNyou need to reboot for it to have effect
16:13:39Rorilet me try reboot
16:14:13Rorithat worked
16:14:26LinusN2) Reset your settings and adjust the wps to the current specification
16:14:47Roriwuh? you mean it's all changed?
16:14:51LinusNthe wps format has changed a bit, especially for bitmaps
16:14:58BgerLinusN iirc there were problems with max files in dir > 5000 or so
16:15:06LinusNBger: really?
16:15:26LinusNi didn't know that
16:15:31Bgermaybe i'm wrong, but i remember something about this
16:15:32Roriplayer crashed again
16:15:50LinusNcrashed how?
16:16:04Rorias in locked up solid with light on
16:16:17LinusNi suggest you reset your settings
16:16:26HClpreglow: my mpg has an internal mp3 stream, so rockbox would be able to play it even though its a video?
16:16:45LinusNHCl: probably not
16:17:12LinusNthe mp3 decoder will bail out if it encounters many sync errors
16:17:12HClwho would i have to talk to to change it so you can play audiostreams from video mpeg2s
16:17:24LinusNHCl: yourself :-)
16:17:26Rorimaybe the wps that now needs redoing is causing the crashes when I try to load them
16:17:40LinusNRori: yes, reset your settings
16:18:32Roriyeah did. I need to fix the wps now....where is the info for changing to the new commands?
16:19:00LinusNin the "documentation" section on
16:19:11LinusNsurprisingly :-)
16:19:28BgerLinusN see , 18.12.47
16:20:08LinusNit is a known problem???
16:20:24Bgeri just remember seeing it ...
16:20:30LinusNwow, i didn't know that :-)
16:20:38RoriOK this is my current fave wps I fiddled with. Any idea what is wrong with it?
16:20:39 Join Mxm`Pas`Bien [0] (
16:20:40rasherIt's one of those known unknowns
16:20:47Rorioh crap can't paste it
16:20:49Bgermaybe it'll be better to ask amiconn for details
16:20:53Bgeri gotta go
16:21:20Rorigimmie a clue on that line and I can adjust from there
16:21:40Bgerbtw, regarding this
16:22:09Bgeri think it's better to search for wps images in the dir where the .wps file is
16:22:35Bgerso you can make /.rockbox/wps1/1.bmp ... 2.bmp
16:22:53 Quit Bger ("BitchX: now in non-drowsy formula too!")
16:23:41LinusNRori: it is no longer 1-9 for images, it's a-z
16:24:06Roriah crap
16:24:44rasherso %x1|wps/edan/e_logo.bmp|116|117| becomes %x|wps/edan/e_logo.bmp|116|117|
16:25:08Rorihmm ok will give it a shot
16:25:13LinusNno, %x|a|wps/edan/e_logo.bmp|116|117|
16:25:30LinusNand %x|b for the next and so on
16:27:02Rorithat works thx
16:27:09Roridamned syntax crapola
16:27:23LinusNsorry for that
16:28:38preglowHCl: you'll need a proper mpeg stream parser
16:29:03preglowthere isn't just one long mp3 stream in the file
16:29:08preglowit's segmented over the entire file
16:30:52LinusNwe *could* adapt the mp3 codec to accept multiple sync errors so it skips the video data and syncs on the audio data
16:31:24LinusNprovided it is segmented on frame boundaries...
16:31:33LinusNwhich i doubt
16:32:14preglowor we could make a demuxer
16:32:18preglowand do the job properly
16:32:30LinusNsure, would be somewhat cool
16:32:38LinusNbut how useful is it?
16:32:43preglownot at all, actually
16:32:46preglowbut would be cool
16:33:24LinusNin fact, one could do a demuxer as a plugin
16:33:34LinusNfor the fun of it
16:34:47preglowit would fit into my scheme of allowing the codecs to load their own data
16:35:09preglowLinusN: what would be cool would be allowing plugins to load a codec
16:35:30preglowLinusN: then we could make a demuxer as a plugin, then just having it load a codec
16:35:53preglowbut this is all talk on my side, i'm not too interested :)
16:35:53LinusNtranscode.rock :-)
16:36:00preglowthat would be nice
16:37:53 Quit Maxime (Read error: 110 (Connection timed out))
16:39:59linuxstb_I've written various MPEG demuxing tools (but mainly for transport streams). But it would have to be done outside the codec - we don't want to waste memory buffering video that we will just be discarding later.
16:40:25linuxstb_But I would very much like to see Rockbox do that.
16:40:28preglowwe'll need to allow codecs to load their own data sooner or later
16:40:31preglowfor streaming codecs
16:40:41preglowand there's nothing wrong in allowing a codec to have several loaders
16:40:50 Join leftright [0] (
16:40:57preglowehh, non-streaming codecs
16:41:25leftrightlinuxstb: how's sudoku coming on ? :)
16:41:40 Join noC|andY`fRa [0] (
16:41:52linuxstb_leftright: It's about ready to commit - I'll have a final clean up when the feature freeze is over.
16:42:04preglowLinusN: is there some way to force a struct to appear as the first thing in a binary image?
16:42:10leftrightcan't wait, love the game
16:42:19preglowLinusN: perhaps stuffing it in a separate section on forcing that first?
16:42:24LinusNpreglow: yes
16:42:31LinusNthat's how
16:42:40preglowsimple way of extending the plugins
16:42:47preglowand still having it be a binary image
16:43:01LinusNor you can force the linker to use a specific section from a specific .o file first
16:43:34leftrightany plans to make the .cfg files more accessable ?, i.e. quick menu ?
16:43:50LinusNno plans
16:44:04leftrightthats a shame,
16:44:13LinusNthat doesn't mean that it won't happen
16:44:26leftrightseeing as cfg files are used frequently to change settings
16:44:31LinusNif a patch pops up, we might just as well commit it
16:44:45preglowlinuxstb_: how hard is demuxing?
16:44:59LinusNin fact, there are very few real plans in rockbox in general
16:46:38 Join Pieter__ [0] (
16:46:54leftrightI need to find a bribeable coder :)
16:47:13Pieter__did anyone else notice the ondio swapping the left and right channels when recording?
16:47:20preglowamiconn did
16:47:35Pieter__just noticed - recorded a piece where there were lightbulbs thrown on the left side
16:47:40LinusNit's a general archos issue
16:47:42Pieter__they appeared on the right when listening back :)
16:47:48Pieter__i'll just swap the channels back
16:48:05LinusNmany archos devices have the line in channels swapped in the connector
16:48:13preglowgood engineering
16:49:01LinusNespecially the recv1, where the connector is soldered by hand with two wires
16:49:21Pieter__guess there's no way to add a 'swap channels when recording' feature? ;)
16:49:28preglownot without hacking the mas
16:49:58Pieter__hehe, thought so.. ok
16:50:53Pieter__i'll just swap the channels so people will stop wondering why they hear the light bulbs on the wrong side :)
16:51:20Pieter__(those things make wonderful noises when thrown at the floor ;))
16:51:37LinusNgotta go, cu all
16:51:39 Part LinusN
16:52:09linuxstb_preglow: Demuxing isn't hard. Seeking in the file will be hard.
16:53:02preglowhow's that handled, then?
16:53:56linuxstb_Like VBR MP3 files I think - i.e. you guess. DVD program streams have extra seek infomation (so-called navigation packets) inserted in them. But "normal" MPEG programs streams don't.
16:54:24preglowwhat were people thinking
16:54:29linuxstb_One advantage of program streams is that audio stream is encapsulated into "PES" (packetized elementary stream) packets - each with a timestamp in the header.
16:56:38linuxstb_The demuxing software I've written has just been for off-line extraction of the elementary streams - not for use in a player application. So I haven't had to worry about seeking.
16:57:21preglowoh well
16:57:26preglowall i want is a wav writer :P
16:59:42Pieter__hmm, audacity doesn't like big recordings :(
16:59:49preglowaudacity doesn't like being used
17:00:00Pieter__no, so it seems
17:00:04Pieter__can't open a 2 hour mp3 file in it
17:00:15 Quit ashridah ("sleep")
17:00:22Pieter__or i can, but it takes an hour, then clicking anything produces nothing but lots of swapping :p
17:02:03 Join webguest46 [0] (
17:04:40preglowaudio editors that can't handle large files without loading them all into "ram" are pretty much unusable
17:08:19Pieter__yeah, so it seems.. i haven't had the problem before, but i've never used this big files
17:08:31Pieter__usually i could at least half way create a new file
17:09:22Pieter__by the way, is there an easy fix for the ondio battery life problem thing? mine gets rather hot and doesn't last very long on it's batteries :(
17:10:23 Part leftright
17:19:25 Quit webguest46 ("CGI:IRC")
17:22:09 Join dpassen1 [0] (n=dpassen1@resnet-233-61.resnet.UMBC.EDU)
17:32:14 Quit Rori ()
17:38:28[IDC]DragonPieter__: only changing the LTC3440 helps
17:39:20amiconn[IDC]Dragon: Still wanting audible ff/rw? I don't care about that feature...
17:39:41preglowi'd love it
17:40:18preglowbut do you think it's possible to find out whether a bit shift has occured in a frame by looking for the sync bits in the mpeg frame?
17:46:05amiconnThe sync bits alone aren't very reliable
17:46:36preglowhow come?
17:46:37amiconnWe'd need to walk the data stream and check all headers
17:47:01amiconnWe know which parameters are fixed and so we know which bits must be fixed
17:47:01preglowyou'd need to check everything coming out of the mas, yes
17:47:14amiconnNo, not everything
17:47:25preglowdepends on when a bit shift error can occur
17:47:50preglowbut the header should suffice
17:48:03amiconnWe just have to search for the first frame header. Then we know how long that frame is and here where the next frame header should occur
17:48:04preglowbut why isn't just the sync bits enough? they're constant
17:48:32amiconnYes, but chances are high you won't notice a single bit shift by just the sync bit
17:49:11[IDC]DragonLuke Framewalker, as I said
17:49:16preglowwhy not?
17:49:22amiconnWell, perhaps it would be reliable enough, but we'll need to find the first header anyway
17:49:26preglowthey'll all be displaced
17:49:40preglowyes, and that's done by checking for the sync bits
17:49:51amiconnWe have an advantage - the mas gives us the last header it decoded/encoded via i2c, so we know what to search for
17:49:58[IDC]Dragonmy rvfmux tool has a frame walker
17:50:36preglowamiconn: yeah, but there's no point in looking for all that when you can save cpu in only looking for sync
17:50:42amiconnpreglow: You can't find a header just by looking for the sync bit sequence. There are sometimes sequences of 0xFF bytes within frames
17:50:59preglowamiconn: that shouldn't be, that's the entire point of the sync bits
17:51:09amiconnIt's definitely the case
17:51:12preglowahh, yes
17:51:17preglowff bytes are ok
17:51:23preglowsync bits are eleven bits
17:51:27amiconnBelieve me, I've looked at megabytes of mp3 data with my hex editor
17:51:32preglownot eight
17:51:41preglowand can be distributed however they like among two bytes
17:52:03preglowas long as they're in sequence, of course
17:52:07amiconnIf no bitshift occured, they are aligned always the same way
17:52:31amiconnHowever, that still doesn't help because sequences of 0xff can occur within frames
17:52:40preglow0xff doesn't matter, that's just eight bits
17:52:54amiconn*sequences* of 0xff
17:53:03preglowthat's extremely weird
17:53:13amiconnIt's not
17:53:15preglowcan't possibly imagine why they bothered with sync bits if that's the case
17:53:32preglowdo you know where in a frame those sequences are located?
17:53:35 Join R-S-W [0] (
17:53:51 Join paugh [0] (n=kickback@2001:5c0:8fff:ffff:8000:0:3e03:6822)
17:54:19amiconnThat's why all mp3 tools that don't know the exact frame header in advance check for several consecutive frames (usually >=3) if they found a maybe-header whether it is really a frame header
17:55:40 Join Sucka`away [0] (n=NNSCRIPT@
17:56:23 Nick Sucka`away is now known as Sucka (n=NNSCRIPT@
17:56:53amiconn[IDC]Dragon: Any ideas how to integrate such a framewalker into the rockbox architecture?
17:56:54preglowwell, that does indeed defeat the entire purpose of the sync bits
17:57:04 Nick Mxm`Pas`Bien is now known as Maxime (
17:57:08 Nick Maxime is now known as Maxime` (
17:57:35amiconnpreglow: I opened the first arbitrary mp3 file, looked at the first frame - there's a 0xff 0xff sequence in it
17:57:47preglowthen you'll indeed need to look for the entire header
17:58:22Pieter__[IDC]Dragon: i can't do that.. oh well :P
18:00:50preglowamiconn: but are you going to try to fix this before 2.5?
18:02:42amiconnI don't think this can be implemented quick enough to justify holding back 2.5
18:04:00amiconnInteresting... waiting for /RTW to become high before entering the loop again made the error occur earlier...
18:04:24preglowamiconn: btw, you sure that 0xff 0xff sequence isn't the sync bit sequence?
18:04:33amiconnYes, definitely
18:04:39preglowamiconn: you'd see a 0xff 0xff sequence in all frames for quite ordinary files
18:04:49amiconnThe sync can't be 0xff 0xff
18:05:00amiconnThe second byte must be != 0xff
18:05:10preglowno, but sync bits + layer 1 + mpeg version 1 + not protected file == the two first bytes of nearly all mpeg files
18:05:22preglowor mpeg frames, i mean
18:05:31amiconnThe mas only records layer 3
18:05:31preglowand that equals 0xff 0xff
18:05:37preglowahh, i mean layer 3, sorry
18:06:21amiconnLayer 3 is 01
18:06:26amiconnLayer 1 is 11
18:06:29preglowyes, quite
18:06:30preglowignore me
18:06:51preglowi know, i'm just confused
18:06:54preglowi'll go attend to my baking again
18:07:21 Quit R-S-W ()
18:08:02amiconnArchos recorded frames always start with 0xfffa or 0xfff2
18:08:08amiconn(we do enable crc protection)
18:08:35preglowbut you don't know which part of the frame you find the 0xff 0xff sequence in?
18:09:04amiconnI don't know much about the internals of a frame. It was around the middle
18:10:52amiconnThis is a 192 kbps frame which is 626 bytes long. 0xffff occurs there at position 382
18:11:07amiconn(position 0 being start of header)
18:11:28*[IDC]Dragon got back
18:11:57[IDC]DragonI can confirm the mpeg audio suffers from so-called start code emulation
18:12:29[IDC]Dragonmeaning, the payload may (by chance) look like a frame header
18:12:48[IDC]Dragon[17:56] <amiconn> [IDC]Dragon: Any ideas how to integrate such a framewalker into the rockbox architecture?
18:12:56***Saving seen data "./dancer.seen"
18:12:58[IDC]Dragonperhaps as a state machine?
18:13:15[IDC]Dragonwe're touching the bytes anyway
18:13:27preglow[IDC]Dragon: is this thanks to bad planning from the mpeg people or bad implementations?
18:13:48[IDC]Dragonthe scpecification allows it
18:16:46amiconn[IDC]Dragon: Where are we touching the recorded bytes? (except for saving them)
18:17:22amiconnAnd how do you mean as a state machine? I'm thinking about the way to integrate it.
18:17:31amiconnShould the mpeg thread do the check?
18:17:40amiconnDo we need a separate thread?
18:18:40amiconnAnd finally, for your beloved feature, the framewalker would need to run for playback as well. That won't work with the current approach of swapping the main buffer
18:19:58[IDC]DragonI know
18:20:03amiconnAudible ff/rw will still be hard even with a frame walker. It's only simple to walk forward, backward is tedious
18:20:21[IDC]Dragonthat, too
18:20:23 Nick solexx_ is now known as solexx (
18:20:24amiconnAnd for ff, it has to be fast enough to cope with the speed
18:20:45[IDC]Dragonwell, Archos does it
18:21:00amiconnYes, and I don't know why (for playback)
18:21:17[IDC]Dragonstop mobbing that feature!
18:21:23amiconnI prefer silent ff/rw over unidentifiable twittering noises
18:21:36preglowhow does archos ffrw compare to irivers?
18:21:42preglowiriver's audible ffrw sounds pretty nice
18:21:50preglowand very nice for vorbis files
18:21:54[IDC]Dragonthe first config option I'm voting for, instead of against
18:22:12amiconnpreglow: I never tried it
18:22:23[IDC]Dragonhow to implement: yes, I ment the byte saving
18:22:31[IDC]Dragon meant
18:23:13amiconnI think we would need to walk directly behind the mas writing. When saving it is too late
18:24:01amiconnPerhaps we could even do that within the transfer isr
18:24:08 Quit bobTHC ("Smoke Weed Every Day !")
18:24:08amiconn(for recording)
18:25:11amiconnIt could also count the frames if it walks them anyway
18:25:45amiconnNo resort to estimation from the recording time if the mas counter saturates
18:26:33[IDC]Dragonyes, that's what I mean, process the header while grabbing the bytes off the hardware
18:26:50[IDC]Dragonbut better have v2.5 released first
18:26:55amiconnThere won't be a header within every DMA block
18:27:12amiconnwe'll need a byte counter for the position within the frame
18:27:13[IDC]Dragonthat's what the state machine is for
18:29:05[IDC]Dragonwe'll have a performance peak during the header, else we're ust counting for the next header
18:29:29amiconnToday's test recording got a +1 bitshift after ~90 minutes. That's with the check for /rtw going inactive before looping
18:31:02[IDC]Dragonhm, my state machine code is not in rvf_mux, but in the DirectShow filter
18:31:16[IDC]Dragonwhich is closed source
18:31:45amiconnIs there a special reason for the directshow filter being closed source?
18:31:50amiconn(just wondering)
18:32:04[IDC]Dragonanyway, it should be much simpler for the MAS
18:32:17[IDC]Dragonyes, too much overlap with my work
18:36:43amiconnI guess that a frame walker could actually simplify many other parts of the recording engine
18:37:02[IDC]Dragonnow we're talking ;-)
18:37:08amiconnToday, we search for frame boundaries when splitting etc
18:37:20[IDC]DragonI gotta leave
18:37:23amiconnThe frame walker would store the last frame boundary position
18:38:30 Quit Bagder (Read error: 104 (Connection reset by peer))
18:38:34[IDC]Dragonthe state machine in essence works like this:
18:38:57[IDC]Dragon- you present it with an arbitrary amount of bytes
18:39:12[IDC]Dragon- the state decides what'll happen next
18:39:38[IDC]Dragone.g. at the init state, you just memchr for an 0xFF
18:39:55[IDC]Dragonwhen found, you're in the next state
18:40:08[IDC]Dragonand verify byte 2 of the header
18:40:30[IDC]Dragonif no match, go back into scan state
18:40:58[IDC]Dragonthen verify byte 3, processing the relevant bits
18:41:25[IDC]Dragonand so on, until the distance to the nect frame is known
18:41:45[IDC]Dragonthen you're in a "skip payload" state
18:41:57[IDC]Dragoncounting off until the next header
18:42:06 Quit linuxstb_ ("CGI:IRC")
18:43:03amiconnWe don't even need to search for the first header at the start of the recording. We know it must be the first thing
18:43:18[IDC]Dragonan extra ontop condition would be to only changed from an "unsynced" to a "synced" state when seen eg. 3 headers with the right spacing
18:43:37amiconn...and if we loose sync, we need to restart recording anyway
18:43:41[IDC]Dragonwell, that was the general case.
18:43:54amiconn...meaning we do again know where our first new header must be
18:43:55[IDC]Dragonsimplifications apply here
18:44:12[IDC]Dragonand we know sample rate, etc
18:44:20[IDC]Dragonno need to extract that
18:44:29amiconnWe don't know the sample rate in case of s/pdif
18:45:14amiconnWhen recording from s/pdif the mas auto-syncs to the input, and ignores the setting
18:45:26[IDC]Dragonperhaps we can assume it being constant
18:45:41amiconnWe know a lot of things, and we can get the header template from the mas itself
18:45:45[IDC]Dragontaking that I2C header
18:45:45amiconn(via i2c)
18:46:14[IDC]Dragonjust for a start, I woudn't use it while running
18:46:28[IDC]Dragonmost likely it is outdated
18:46:28amiconnWhy not?
18:46:37amiconnYes, it's only a template
18:46:45[IDC]Dragonwe don'tknow which frame it really belongs to
18:46:56amiconnbitrate and mode extension can change
18:47:19[IDC]Dragonbut then a restart probably woudn't hurt
18:47:39[IDC]Dragonbitrate changes all the time
18:47:58[IDC]DragonI rather meant sample rate
18:48:13amiconnVersion, layer, protection, channel modecopyright, original bit, emphasis, padding and private bit are fixed
18:48:36amiconnSample rate should be fixed too
18:48:43amiconn(within one recording)
18:49:00amiconnpadding is simple - *all* mas encoded frames are unpadded
18:49:39[IDC]DragonV said the MAS has an undocumented feature, doing CBR
18:49:56[IDC]Dragonbut in a dumb way: it pads like crazy
18:50:29amiconnYes, I read that note about the AV3x0 firmware update enabling cbr
18:50:43amiconnI wonder how this mode is enabled...
18:50:45[IDC]Dragona waste of diskspace
18:51:11*[IDC]Dragon really gotta leave
18:51:25 Quit [IDC]Dragon ("CGI:IRC")
19:22:03 Quit dpassen1 (Read error: 110 (Connection timed out))
19:23:08 Join Dan [0] (
19:23:42Danis there a max volume setting on rockbox?
19:24:08Danbecause rockbox is a hell of allot quieter than the iriver firmware4
19:30:40crwljust turn bass / treble down, rockbox doesn't limit them like iriver firmware does
19:32:35Danstill really quiet
19:46:36 Join Lear [0] (
19:48:39 Quit markun ()
20:01:04 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
20:12:52preglowDan: replaygain on? _completely_ certain bass and treble is zero?
20:12:58***Saving seen data "./dancer.seen"
20:19:23 Quit Dan (Read error: 110 (Connection timed out))
20:31:50 Part nibbler ("Leaving")
20:33:09 Join tomal [0] (
20:35:50 Join solexx_ [0] (
20:41:07 Join dpassen1 [0] (n=dpassen1@resnet-233-61.resnet.UMBC.EDU)
20:42:52 Join Seed [0] (
20:47:50 Quit solexx (Read error: 110 (Connection timed out))
20:58:20 Join Yono [0] (
20:59:08 Join matsl [0] (
21:12:41 Quit tomal ("leaving")
21:22:10 Quit Maxime` ()
21:25:13 Join Maxime [0] (
21:32:41 Quit tvelocity ("Leaving")
21:34:30 Quit Yono ("Download Gaim:")
21:46:04 Quit t0mas (Read error: 104 (Connection reset by peer))
21:46:18 Join t0mas [0] (n=Tomas@unaffiliated/t0mas)
21:47:04 Quit t0mas (Read error: 104 (Connection reset by peer))
21:55:50 Join t0mas [0] (n=Tomas@unaffiliated/t0mas)
22:01:53 Quit t0mas (Read error: 104 (Connection reset by peer))
22:13:02***Saving seen data "./dancer.seen"
22:14:03 Quit Lear ("Chatzilla [Firefox 1.4/undefined]")
22:25:34 Join travis [0] (
22:25:42travisfreenode-connect is being nosey!
22:26:13 Quit travis (Remote closed the connection)
22:27:01 Join travis95 [0] (
22:27:05travis95freenode-connect is being nosey!
22:27:52travis95Zagor is being nosey!
22:28:06Zagorwith all right, since you are acting like a bot
22:31:08 Quit travis95 (Client Quit)
22:56:36 Quit dpassen1 ()
22:59:17 Quit matsl (Read error: 104 (Connection reset by peer))
23:00:51 Join matsl [0] (
23:16:47 Join unL33T [0] (
23:17:10 Quit unL33T (Client Quit)
23:17:22 Join Moos [0] (
23:17:24 Join unL33T2 [0] (
23:18:22unL33T2hey i have a problem when trying to use the regular iRiver firmware on my iRiver HP-120 ... anyone wanna try to help me with it =)
23:19:43unL33T2basically if I select a song through the artist or album or one of the other sorted menus it just doesn't play it and just closes the menu and goes back to whatever was there before but if I go through the folder browser menu anything I select works fine.... I've tried rebuilding the database with MoodLogic a couple times ... last I'm thinking is a reformat of the player but I'd rather not as it is also my primary music storage
23:20:23unL33T2rockbox still works normally but I'm still not sure if I like it better than the regular firmware so I'm still switching between both
23:21:24Zagori think there was a bug in the bootloader causing something like that. are you using the latest version?
23:23:30unL33T2I downloaded the 1.65 firmware from the links provided in the faq and ran the patching program ... unless the bootloader is something other than the hacked firmware...
23:23:47unL33T2I'm new to using so kind of dumb
23:24:08unL33T2or does the bootloader depend on the version of rockbox I have?
23:25:03Zagorno. but the bootloader has been updated since the bug was found. try grabbing the 1.65 firmware and patch it again.
23:25:32amiconnI wonder whether it would be a good idea to display the bootloader version within rockbox
23:25:44amiconn(somewhere in the debug menu)
23:25:55Zagormight be a good idea, yes
23:26:00unL33T2i just downloaded and patched it two days ago
23:26:12ZagorunL33T2: ok, doesn't sound like that then
23:26:35amiconnThere's an orphaned "View HW info" menu item in the iriver debug menu...
23:27:01unL33T2and what should that tell me?
23:27:19amiconnThat was directed to Zagor (and other devs)
23:27:29unL33T2oh sorry :P
23:28:10unL33T2hmm I just noticed that when I turn on my iriver it says H-100 series (which I think it always said) but the player is an iHP-120 ... I dunno if that is significant.
23:30:39Zagoryou didn't use the wrong model firmware, did you?
23:31:11unL33T2positive I didnt but I'll verify the filename
23:31:49Zagoroh, my 140 says "H-100 series" too, so don't worry
23:32:27unL33T2I've tried reflashing to the stock firmware as well ... both 1.63 and 1.65 and have the same problem
23:32:50unL33T2which is why i suspect that it is a problem with the database but I dont know how to fix it
23:32:53preglowSeed: still no news as to fixing libmpcdec for rockbox, i imagine? :P
23:33:26amiconniriver firmware is dumb... it told me "low battery" when I wanted to check what it says on boot. Rockbox still runs happily...
23:33:51unL33T2all my firmwares I have are all ihp_120.hex
23:34:05preglowwell, iriver firmware has a safety check to avoid angry users...
23:34:14*amiconn wonders what this mpeg frame checker is doing with his ram...
23:34:16preglowwhich is perfectly understandable
23:35:11preglowunL33T2: what version bootloader do you have? it says so first thing when you turn the unit on
23:35:24amiconnpreglow: I'd rather like to use the full capacity of the battery...
23:35:51unL33T2"rockboot version 2"
23:36:22unL33T2is there a way to pause it during boot so I can see all that info it provides?
23:37:55amiconnOh, that's old... current bootloader is v5.
23:38:04amiconnHmm, the wiki contains v2 - argh!
23:38:04 Quit noC|andY`fRa (Read error: 104 (Connection reset by peer))
23:38:44unL33T2that is probably where I got it
23:38:48rasherI have v5.. hang on
23:39:00unL33T2ok thanks
23:39:22unL33T2is it the same process to install it? just patch the firmware with a newer program?
23:40:04 Join DrMoos [0] (
23:40:09 Quit Moos (Read error: 104 (Connection reset by peer))
23:40:15preglowamiconn: sure, but i bet iriver doesn't trust their ordinary userbase to do so
23:40:29preglowoh shit
23:40:31preglowi didn't think of that
23:40:35 Nick DrMoos is now known as Moos (
23:40:45amiconnpreglow: ?
23:40:50preglowwiki and old fwpatcher
23:40:58preglowwhich is kind of nasty
23:41:26amiconnObviously google didn't like the IriverBoot page
23:42:10preglowv5 should work perfectly for you
23:42:28unL33T2cool I just have to get it =)
23:43:13rasherHrm, is the last one "20050722?
23:43:44preglowgimme a sec
23:44:09rasherHrm.. how do I delete an attachment again?
23:44:46preglowwhat the hell?
23:44:53preglowsuddenly the bootloader uses 10 seconds to get going
23:44:58preglowi have to stop booting the iriver fw
23:45:30 Join ashridah [0] (
23:46:09amiconnrasher: By moving it to Trash/TrashAttachment, but why do you want to delete it?
23:46:22rasherwell, what do I call the bootloader.bins ?
23:46:27rasherh120 / h100
23:46:41rasherccurrently there's just "bootloader.bin"
23:47:35 Join DrMoos [0] (
23:47:56amiconnAh, that problem...
23:48:24 Quit Moos (Read error: 104 (Connection reset by peer))
23:50:41rasherWhat do I do?
23:53:13rasherI haaaave the file... where to put it?
23:56:17unL33T2I dont even know what .bin files are used for ....
23:56:27rasherfwpatcher.exe updated
23:57:39unL33T2so this link should be to the correct one: ?
23:57:40rasherthink I'll leave it to Linus to upload the .bins

Previous day | Next day