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 2005-02-05

00:00:41 Quit Remo ("ChatZilla 0.9.61 [Mozilla rv:1.7.3/20040910]")
00:07:35 Quit Nibbler (Read error: 110 (Connection timed out))
00:08:42preglowBagder: you should change your nick for simplicity
00:10:16BagderI've used this nick for almost 20 years now
00:10:24amiconnCassandra_: My comments on your patch: (1) I have to agree with Bagder, this mixes firmware and app level (started_from_rtc variable)
00:10:31preglowso you agree it's time for change? :P
00:10:57amiconn(2) The alarm flag is read only, you can't reset it by writing the data back with the flag reset.
00:12:51Cassandra_The first I fixed by making rtc_init return the value.
00:13:04Cassandra_I'll take out the redundant code in that case.
00:13:05Bagder-rw-r−−r−− 1 daniel daniel 104064 Feb 5 00:12 libmad.a
00:13:12Bagderbuilt with m68k
00:13:15preglowwould you care to share the procedure?
00:13:36Bagderhang on
00:13:55Bagder(I haven't linked with it yet)
00:14:07preglownono, but just building a lib is more than i managed, hehe
00:15:34Cassandra_Right, suppose I'd better test this again, just to make sure it still works.
00:16:05amiconnCassandra_: If you return the value from rtc_init(), how do you store it? It is needed later, and in another place than the rtc_init() call takes place.
00:16:25Cassandra_In a global bool.
00:16:31amiconnI checked the datasheet, the alarm flag resets itself as soon as the flags register is read.
00:16:45Cassandra_That's good.
00:17:11preglowbut yes, now for supporting the emac stuff, which will be more than enough fun, since gcc almost certainly doesn't know about the four accumulator registers
00:17:34Bagderno it doesn't
00:17:51amiconnI think a separate function for checking the flag would be more straightforward than a global bool. Plus, since the flag resets itself after being read once, there's no need to reset a separate variable.
00:18:30 Join Nibbler [0] (
00:18:40amiconnCassandra_: Btw, your work is the first step towards a recording timer...
00:19:08Cassandra_True - cool
00:20:29Cassandra_Actually I think that's fairly trivially implementable in combination with the "start in recording screen" option.
00:21:35amiconnThe function to check for powerup from alarm can be written as a one-liner:
00:21:50amiconnbool started_from_alarm(void)
00:21:50amiconn return (rtc_read(0x0f) & 0x40) != 0;
00:21:50DBUGEnqueued KICK amiconn
00:22:44Cassandra_You're allowed to call rtc_ functions from the app, right.
00:22:51Cassandra_Well, of course you are, doh.
00:23:10Cassandra_OK. Means you don't need the global var either.
00:23:32Cassandra_Excuse me while I go implement starting recording from RTC alarm.
00:24:11Bagderpreglow: I tried 'make install' too after that command and it runs fine, making it easier to build rockbox with it later
00:28:17amiconnCassandra_: "Start in recording screen" + "wakeup alarm" => timer recording
00:30:38Cassandra_Just need to add a little patch to the recording screen code.
00:34:43amiconnBagder: What means: "make[2]: *** No rule to make target `<<<<<<<', needed by `/home/Administrator/rb-patched/build/fmrecorder/dep-plugins'. Stop." ?
00:35:11amiconnThis happens when I "make clean"
00:35:26Bagderremove the dep-plugins file
00:35:43Cassandra_Damn - I need the state twice. Better to use a static in rtc_check_alarm_started to remember state, or use a global to hold it once it's been read?
00:36:10amiconnWhy do you need it twice?
00:36:27Cassandra_Once to make sure I don't resume, once to start recording.
00:37:37Cassandra_Once to make sure I don't ask the user whether to resume , once to start recording without user intervention.
00:37:42amiconnI thought your start-in-recording-screen first checks whether recording screen should be started or normal browser+resume check
00:38:23amiconnYes, but you only need either one or the other check per each boot, never both
00:38:25Cassandra_Nope, resume can override start in recording screen.
00:38:33Cassandra_That seemed like the right way round to do it.
00:39:17amiconnReally? That way, the "start in recording screen" wouldn't do anything good for me in >90% of all cases.
00:39:40Cassandra_You have resume = yes?
00:40:07amiconnNo, I have resume = ask, but I practically always have a stored resume position.
00:40:36Cassandra_If you say "no" to resume, it drops into the recording screen.
00:41:37amiconnAh ok. Still, for timer recording, this wouldn't do anything useful, because with you resume-from-alarm patch it would always resume playback without asking you
00:41:52Cassandra_I thought it best not to have "start in recording screen" disable resume functionality.
00:42:43amiconnIt would have to if started from alarm and "start in recording screen" is on.
00:43:33Cassandra_Well yes, resume functionality is overriden when starting from alarm.
00:43:48amiconnYes, that's exactly what I mean.
00:43:53Cassandra_But in order to do that, you have to check to see whether you're starting from alarm.
00:44:22Cassandra_So you need two checks - once to avoid the resume code, and once to start recording.
00:44:23amiconnYou can still do it the other way round.
00:45:05amiconnOn boot, first check whether "start in recording screen" is enabled
00:45:06Cassandra_I don't see how that would help.
00:45:49Cassandra_Oh, I see, yes. Good plan.
00:45:54amiconnIf yes, then check for startup-from alarm, and if again yes, start the recording screen with a parameter that tells it to start recording immediately
00:46:45amiconnIf the latter says no, then continue with normal resume, and if no resume, then call recording screen
00:47:14Cassandra_Ah, that'd involve calling the recording screen from two different points in the code.
00:47:33Cassandra_I could just be being dense, but I think it'd screw up the execution flow.
00:48:46amiconnIf you do it the other way, i.e. first call the resume check, the resume check would need to take the start-in-recording-screen setting into account, which is also not very logical
00:49:47***Saving seen data "./dancer.seen"
00:49:56Cassandra_Sorry, I really am lost now. I don't think so, but I'm too confused to work it out. Sorry.
00:50:25Cassandra_The program flow around startup is pretty nightmarish.
00:51:03amiconnThat's because the resume check would need to differentiate 4 conditions
00:51:25amiconn(1) Not started from alarm, and not start-in-rec-screen: Normal resume request.
00:51:33Cassandra_At the moment, the only logic I need in the resume check is whether or not it's started from rtc.
00:51:53Cassandra_Oh, hold on. I think I'm getting ther.
00:52:01amiconn(2) Started from alarm, not start-in-rec-screen: Resume unconditionally.
00:52:36amiconn(3) Not started from alarm, start-in-rec-screen: Normal resume request, if not resumed, drop to recording screen
00:52:39CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
00:52:39*Cassandra_ nods.
00:53:01amiconn(4) Start from alarm, start-in-rec-screen: Start recording screen & recording immediately.
00:53:19Cassandra_So you'd suggest changing the logic so that "start in record" overrides resume?
00:53:19Quel|awayif i can give my opinion, i'm with amiconn
00:53:36Cassandra_Yeah, he's right. I was just being dense.
00:53:37amiconnYou see, you'll always have two places from where to call the recording screen
00:54:11amiconnANd you'll always have to check started-from alarm only once.
00:54:40amiconn(using a 4-way if-else construct for the 4 possibilities mentioned above)
00:55:08amiconnThere are different ways to implement this though
00:55:22Cassandra_Yeah. I think I'll leave it for tonight.
00:55:35Cassandra_I need to think more about it.
00:56:36amiconnAnd: I have to agree, the startup is nightmarish.
00:56:57Quel|awayi suppose you have to check first of all a *start from alarm* and later on *start in rec screen*
00:57:15Quel|awayjust my opinion
00:57:22amiconnThe order of these checks doesn't matter
00:57:52Quel|awayof course
00:58:03Quel|awaybut i like it that way :D
00:58:05amiconnIf you do it the other way round, you'll have 2 places where the check function is called, but only one would get executed per boot
00:59:18cYmenwhich jukeboxes have a built in micro again?
00:59:44amiconncYmen: All recorders have that
01:00:09amiconn(Recorder v1, v2, FM recorder, and Ondio FM)
01:00:54cYmeni only used it to listen to music and as portable disk but rockbox has a nice dictation function hasnt it?
01:01:24Quel|awaydictation function?
01:02:03cYmenso i can use it as dictating machine
01:02:53Cassandra_Right - bed.
01:02:55Cassandra_Night all
01:03:04Quel|awayi think i don't understand that in english... you mean recording your voice for a later use? or what?
01:03:09 Part jyp ("poof!")
01:03:09 Quit Cassandra_ ("Client Exiting")
01:03:12cYmenQuel|away: yes
01:03:20Quel|awaythen, it has
01:03:38cYmenQuel|away: what is your mother tongue?
01:04:41cYmenvoy aprendo espanol
01:04:55Quel|awaymore or less :)
01:04:55preglowsoy el diablo
01:05:05preglowbed for me too
01:05:11Quel|awaynite preglow
01:05:16 Quit preglow ("arf")
01:05:28lolo-laptopprefiero espanol, pero no sabe mucho :-P
01:05:46Quel|awayi'm going to propose spanish as second language for rockbox. Everyone here seems to know some spanish :)
01:06:07lolo-laptopmuy bien!
01:06:54*rasher hides under a rock
01:09:39lolo-laptoppor que el no habla el espanol?
01:10:00*lolo-laptop causes trouble
01:10:15*rasher explodes
01:11:50 Join bagawk [0] (~Lee@bagawk.user)
01:14:58 Quit bagawk (Client Quit)
01:21:53cYmenlolo-laptop: prefiera el espanol a que? (correct me when i make mistakes!)
01:23:47lolo-laptopa ingles, por que ingles es muy loco. Yo no se bastante espanol a hace coreciones... (not sure on some of that)
01:26:50cYmenwell... let s stick to english ;P
01:27:17lolo-laptopgood idea
01:34:19 Quit lolo-laptop ("Client exiting")
01:35:14Quel|awayi think i have to correct both of you
01:35:43cYmengo ahead :)
01:35:53Quel|awayyou just made a small mistake
01:36:23Quel|awayits "Prefiere" or "Prefieres", first one is Polite and the second one is coloquial
01:37:03Quel|awayand the answer... well..i don't know what did he want to say :) sounded more or less ok, but... not spanish ;)
01:37:33cYmenwell i know what he wanted to say :)
01:38:10Quel|awaydid he wanted to say that english is too crazy?
01:38:49cYmenyes and that he doesn t know spanish well enough to correct me :)
01:39:23Quel|awaythen that was ok
01:39:42Quel|awaythe second part is "No se bastante espa˝ol para hacer correcciones"
01:39:49Quel|awaybut was ok
01:40:00Quel|awayeveryone could understand him
01:40:10Quel|away(as you do when i speak english)
01:40:25cYmenenglish is pretty simple and pretty variable
01:40:33cYmenwhatever you say seems to be right in some way :)
01:50:03 Join sdbarker [0] (
01:52:23sdbarkerI just got my new xs200 as an upgrade from recorder15.
01:57:40sdbarkerAlright, so, anyway, who do I talk to about getting on the dev team to work on porting rockbox to this thing?
02:00:19cYmenfirst you talk to the docs then to some tutorials and when you ve become part of the dev team - to yourself :)
02:00:42sdbarkerI should have worded that differently...
02:00:58sdbarkerI read in one of the email list threads, which I can't find now, that somebody had already started on it.
02:01:06cYmenyeah it sounded a bit demanding ;)
02:01:10sdbarkerInstead of duplicate efforts, I'd rather contribute.
02:01:34sdbarkerSo, does anybody know off hand who that is?
02:07:40dwihnoanyone got the iriver display width/height handy?
02:08:56Quel|awayhi dwihno!
02:09:06Quel|awaynice to see you awake
02:09:14dwihnobueno quel
02:09:34Quel|awaymur da bra?
02:10:17rasherdwihno: 160x128
02:11:27 Join lostlogic [0] (
02:11:30dwihnorasher: thanks
02:11:52dwihnoI'll prep some stuff for the target.
02:14:30rasherIf nothing else, I'll sit and answer questions by learning where stuff hides in the wiki :)
02:14:48rasherI guess that'll be my contribution for now
02:15:06 Part sdbarker
02:15:57 Join edx_ [0] (
02:21:12*lostlogic cvs co's rockbox, time to see what people are talking about.
02:26:20 Quit edx (Read error: 110 (Connection timed out))
02:27:09 Nick edx_ is now known as edx (
02:27:47 Join Patr3ck [0] (
02:28:29 Part amiconn
02:40:04lostlogicbah, just took me 3 tries to get my change in the wiki adding more info about the Vorbis codec right. why is it and not org?
02:43:12zewhy is it oggenc and not vorbisenc and vorbiscomment and not oggcomment
02:43:48lostlogicze: to give you something to ask questions about.
02:43:57zethem vorbis folk seem to make some odd choices
02:44:20lostlogicze: hehe, it makes little sense to me, because supposedly ogg is a container (has comments and streams) and vorbis is a stream (needs encoding)
02:44:33zevorbis is the actual audio codec
02:44:49zeogg can contain any kind of media, whatever audio or video codec
02:44:56zesorta like avi can
02:44:56 Join ashridah [0] (
02:44:59zebut better
02:45:05lostlogicor so they say
02:45:33zethere's divx rips out there in ogg
02:45:42lostlogicooh, I'd like to see that :)
02:45:55zealthough a bunch of misguided people have been naming video oggs .ogm
02:46:15zeto distinguish them from audio .ogg's
02:46:24zewhich does make some sense but *shrug*
02:46:29ashridahze: .ogm is a container formate similar to .avi last i heard
02:46:35ashridah.ogg is an audio codec
02:46:40lostlogicashridah: no.
02:46:41lostlogicyou are wrong
02:46:42zevorbis is the audio codec
02:46:45lostlogicyou missed the discussion above
02:46:57*ashridah colours himself corrected
02:47:26zebut yeah are the comments held by the ogg file? i would imagine
02:47:36zeor are they part of the codec (would be strange)
02:47:39lostlogicthat's why the codec as listed on the rockbox wiki is 'OGG/Vorbis' to indicate ogg files containing vorbis streams... kinda like GNU/Linux or something of the sort but different
02:47:52lostlogicze: hmm, lemme take a look
02:48:22zecause ya know if the ogg container held the comments, then 'oggenc' and 'vorbiscomment' are really exactly backwards
02:48:53zewhich, i dunno, maybe there's something intentional to that
02:49:48***Saving seen data "./dancer.seen"
02:50:19lostlogicafaics, the ogg container holds a comment section and then a vorbis stream section (which is prefixed amazingly by the ascii characters v o r b i s :-D
02:50:46zeso maybe the whole thing is a part of a greater backwards theme
02:51:08Quel|awaybye! is the parent 'company' and is the opensource portion?
02:51:12zeinterestingly, seems to actually be registered by the same owners of and has the same name servers listed
02:51:15 Quit Quel|away ("KVIrc 'Realia'")
02:51:35zeand yet has nome A record
02:51:49zedunno where that me came from
02:51:49lostlogicyeah, which is why I had to change it when I initially had it wrong on the wiki
02:51:52ashridahprobably to protect against domain squatters
02:52:05zeyeah but why not use at all?
02:52:13zei mean not even point it to the same place as
02:52:26zeya know google's got and and they both go the same place
02:52:31zeand theres several others like that
02:52:39ashridahgood question. maybe to distance themselves from the idea that they're an organisation, not a company?
02:54:05zei think maybe lostlogic is right about .org parent company and .com opensource
02:54:19lostlogicI just think they're nuts.
02:54:20zeand that going along with some sorta backwards theme
02:55:30zespeaking of xiph
02:55:57zewhatever happened to the cdparanoia that could extract audio from those horribly mutilated CD's that they had a gallery of
02:56:46lostlogicze: was this a special version of cdparanoia? because I've had no problems with my pretty ugly CDs with cdparanoia as it is
02:57:13 Quit mecraw ()
02:58:05zewell i'm more disappointed with that impressive gallery being no more
02:58:08zeat least as far as i can find
02:58:38zeit seemed like the more recent versions had more trouble with some of my cd's than the older one that accompanied those pics did, but that could be entirely unfounded
02:59:03zeall the docs seem to imply that the newer is better
03:05:52*lostlogic is loving the tech docs for rockbox
03:07:04 Join piratePenguin [0] (
03:12:12lostlogichmm... now I hafta figure out how to build a couple cross-compilers on Gentoo
03:17:15lostlogicgrr, the style documentation says "no tab characters" but the source is full of the gawdamn things
03:22:53zetr '\t' ' '
03:23:36lostlogicchuckle - in vim - :set tabstop=4 OR :%s/\t/ /g
03:23:59zei was gonna suggest that 2nd one too
03:24:08zebut i didn't really think you'd be complaining about tabs if you were using vim
03:24:35lostlogicI default tabstop to 100 so that vim autoindents with spaces... because I'm too dumb to figure out the right way to do that
03:25:02zei never do anything with tabstop at all
03:25:07zei just let tabs do what they do by default
03:31:31 Quit Patr3ck ("User pushed the X - because it's Xtra, baby")
03:43:31 Quit Stryke` ("Friends don't let friends listen to Anti-Flag")
03:44:07lostlogicack, bloody brackets on their own lines coding style :-P
03:55:43 Part piratePenguin
04:01:30*ashridah huggles inflated linecounts ;)
04:01:59lostlogicI use a pretty little sed line to strip such bs when I do LOC counts
04:06:18lostlogicI lied, it isn't pretty at all.
04:08:35lostlogiclostlogic@lost ~/rockbox $ ./
04:10:03 Quit QT (Read error: 60 (Operation timed out))
04:11:21lostlogichmm... that would leave in multiline cstyle comments that start on a line with real code, but continue on non-code lines... I'll have to fix that at some point.
04:26:06rasheror you could use sloccount :)
04:28:22lostlogichmm... I suppose
04:28:33lostlogicsed and grep are fun though
04:28:37lostlogicand it's good to practice REs
04:33:58lostlogicI think sloccount overcounts
04:34:04lostlogicby about 5k on rockbox
04:34:24lostlogichmm... maybe it's just counting files I was ignoring like makefiles
04:34:27lostlogicfun times
04:36:54*rasher wanders off
04:44:32lostlogicsweet, there is a gentoo cross compiling howto, m68k-elf here I go.
04:49:49***Saving seen data "./dancer.seen"
05:05:56 Quit cYmen ("leaving")
06:48:13 Join piratePenguin [0] (
06:49:50***Saving seen data "./dancer.seen"
07:06:19 Quit MooMaunder (Read error: 110 (Connection timed out))
07:13:33 Part piratePenguin
07:14:52 Join midk [0] (
07:18:10 Join Trevmar [0] (
07:18:19 Quit XShocK (" HydraIRC -> <- Go on, try it!")
08:29:51 Quit midk (Read error: 60 (Operation timed out))
08:47:17 Quit Trevmar (Read error: 110 (Connection timed out))
08:49:53***Saving seen data "./dancer.seen"
09:00:16 Join SoulEata [0] (
09:00:27SoulEatawow this is amazing.
09:00:48SoulEatafor some reason my college firewall lets me connect to dalnet but here I am on another IRC network. this is exciting.
09:00:55SoulEataonly dalnet*
09:03:54 Join methangas [0] (
09:36:35 Join amiconn [0] (
09:52:02 Nick hile__ is now known as hile (
09:53:15 Join QT [0] (
10:49:56***No seen item changed, no save performed.
12:00:27 Join cYmen [0] (
12:18:49 Quit edx (Read error: 110 (Connection timed out))
12:49:57***Saving seen data "./dancer.seen"
12:56:03 Join jyp [0] (
14:10:01 Quit amiconn (" updating client...")
14:12:22 Join Quelsaruk [0] (~kvirc@
14:14:07 Join amiconn [0] (
14:24:31 Quit Nibbler (Read error: 60 (Operation timed out))
14:29:54 Join Nibbler [0] (
14:43:00 Join preglow [0] (
14:49:59***Saving seen data "./dancer.seen"
15:11:11 Quit methangas (" HydraIRC -> <- The future of IRC")
15:22:54 Part amiconn
15:34:07 Join cYmen_ [0] (
15:35:21 Quit cYmen (Read error: 104 (Connection reset by peer))
15:56:07preglowhas audiomulch-ross been around lately?
15:56:12preglowwrong channel
16:04:02 Join lImbus [0] (
16:04:11lImbusuh. hi all
16:07:15lImbusbuenos dias :)
16:08:15Quelsarukbuenas tardes
16:08:19Quelsarukat least here
16:08:45lImbusoh well. tardes must mean something like "late"
16:09:28Quelsarukand afternoon
16:09:51lImbusi see
16:10:01Quelsaruktarde is late, but also is how we say to evening and afternoon.
16:10:30lImbusthats different from french
16:13:11Quelsaruki think so
16:13:14Quelsarukare you french?
16:13:31 Quit Nibbler (Remote closed the connection)
16:13:57lImbusI am belgian, so implicit french speaking, working in germany, so implicit german speaking, english speaking by business
16:14:31 Join DMJC-L [0] (
16:14:59QuelsarukMaye you could help with French language file...
16:15:10Quelsarukour french expert is missing :(
16:15:37lImbusI can try
16:15:46lImbus(dcc does not work for me)
16:16:03Quelsarukyou can submit to sourceforge :)
16:17:56lImbuswhat do you expect for me to do now ?
16:18:19lImbusdo you have a special problem, or should I just go and check for todo in the file ?
16:20:00 Join Nibbler [0] (
16:21:11 Quit Nibbler (Client Quit)
16:23:04 Join Nibbler [0] (
16:23:09 Quit Nibbler (Read error: 104 (Connection reset by peer))
16:23:46 Join Nibbler [0] (
16:24:42 Join R3nTiL [0] (~zorroz@
16:24:58lImbusQuelsaruk, are you there ? I see a lot of diff between english.lang and francais.lang, but I'm not sure with all of those
16:25:08lImbusfurthermore, I can't submit to cvs
16:25:21lImbusso I need you to commit
16:25:55Quelsaruki can submit, don't worry
16:26:06lImbus1. Question: in english.lang there are a few string around line 430 marked deprecated
16:31:22 Quit ashridah ("sleep")
16:33:06 Quit Nibbler ("blubber")
16:33:41 Join Nibbler [0] (
16:45:16 Quit R3nTiL ()
16:50:02***Saving seen data "./dancer.seen"
17:07:20 Quit lImbus (" client restart")
17:07:27 Join lImbus [0] (
17:12:23 Join XShocK [0] (
17:14:13 Quit XShocK (Client Quit)
17:17:11 Join Cassandra [0] (~christi@
17:17:51 Join XShocK [0] (
17:27:18 Nick gromit` is now known as maison`de`retrai (~gromit`
17:27:25 Nick maison`de`retrai is now known as maison2retraite (~gromit`
17:28:39 Nick maison2retraite is now known as gromit` (~gromit`
17:39:36 Quit XShocK (Read error: 104 (Connection reset by peer))
17:41:24jypWhen compiling locally, compilation stops on 'lang.h: No such file or directory'
17:41:45jypI guess this is a misconfiguration
17:41:55jypcan anyone tell me how to fix this ?
17:44:33 Quit DMJC-L (Read error: 110 (Connection timed out))
17:44:51Quelsarukthe fast way is to use Genlang
17:45:03Quelsarukthat script will create a lang.c and lang.h
17:45:15Quelsarukbut when compiling that is done automatically
17:47:25jypI guess I did something stupid when hacking the Makefiles in my early days ;)
17:47:45jypI'll give it a fresh start
17:59:35 Join Eodun [0] (
17:59:42Quelsarukhi Eodun
17:59:50Eodunhi quel / all
18:00:03EodunI've seen the new langs
18:00:17Quelsarukdo you like the translation?
18:00:27Eodunhaven't dl yet
18:00:31Eodunmy exams...
18:00:32Quelsarukonly a new string
18:00:35Quelsaruki know...
18:00:38Quelsarukuniv is hard
18:00:55Eodunoh... did you modify LANG_DIR string?
18:01:16Eodunit was still untranslated
18:01:34EodunI noticed when I put new music in my box
18:02:40Quelsarukdid i?
18:02:45Quelsaruki can't remember
18:03:39Quelsaruklet me check
18:04:21Eodunnot sure if it was LANG or VOICE_dir, really
18:07:24Quelsaruki'll check when i have a spare moment (bit busy right now)
18:08:33 Join XShocK [0] (
18:13:20 Quit Eodun ()
18:50:06***Saving seen data "./dancer.seen"
19:07:00 Nick Quelsaruk is now known as quel|out (~kvirc@
19:13:28XShocKdid somebody test the bootloader besides Linus?
19:17:55 Join Stryke` [0] (
19:50:19 Join Marder [0] (
19:55:33 Join El_Barto2 [0] (
19:59:40El_Barto2do you know how i could see if my player is charging the batteries▀
19:59:46El_Barto2jukebox recorder..
19:59:52El_Barto2is there a led?
20:11:20quel|outif the recorder is charging?
20:11:25quel|outusing rockbox?
20:11:39quel|outcheck the batt graphic in the top of the screen
20:11:47quel|outyou should see it animated
20:11:57 Join Bluechip [0] (
20:12:11quel|outor, you can go to Main Menu (F1)−−> Info −−>rockbox Info
20:12:29quel|outand check wheter the batt: show a level, or Charging
20:16:30 Part Bluechip
20:17:06 Join Eodun [0] (
20:17:35Eodunhi again
20:27:20 Join quelsaruk [0] (~kvirc@
20:28:41El_Barto2the problem is...
20:28:53El_Barto2the soft hangs
20:28:59El_Barto2and i can do nothing
20:35:12quelsaruki lost the conversation
20:35:37 Join jorbond [0] (
20:44:45 Join thegeek [0] (
20:45:50 Quit quel|out (Read error: 110 (Connection timed out))
20:46:10El_Barto2no prob
20:48:49 Join lImbus924 [0] (
20:50:07***Saving seen data "./dancer.seen"
20:50:47 Quit lImbus (" I love my HydraIRC -> <-")
20:51:01 Nick lImbus924 is now known as lImbus (
20:51:28 Part Marder
20:58:06quelsaruklImbus: do you see it?
21:00:28 Quit cYmen_ ("leaving")
21:00:52 Join cYmen [0] (
21:09:47quelsarukmust go
21:09:56 Quit quelsaruk ("KVIrc 'Realia'")
21:11:03 Quit Eodun ()
21:26:34 Quit jorbond (Read error: 110 (Connection timed out))
21:26:34 Quit thegeek (Read error: 104 (Connection reset by peer))
22:09:31 Quit El_Barto2 ()
22:47:47 Join Patr3ck [0] (
22:50:10***Saving seen data "./dancer.seen"
22:52:14Bagderit feels like a good evening for some good commits! ;-)
22:53:29Bagderit seems Rockbox is almost starting the dir browser fine on the iriver
23:02:31preglowhow is the bootloader testing coming along, btw? still only linus and his h120, no?
23:03:16Bagderbut I hope for some progress in his end
23:03:36preglowwhat still needs progress?
23:04:17BagderLinus said he tried booting (the real) rockbox
23:04:32Bagderand it showed the logo, the disk spun and
23:04:35preglowyeah, and it crashed, but this shouldn't depend on linus anymore?
23:04:36Bagderthen hung
23:04:47preglowor might this be a bootloader thing?
23:04:48Bagderit is probably an init issue
23:05:11preglowyes, it might of course be that
23:05:18preglowis all device init put in the bootloader?
23:05:32Bagderno, it is re-inited by rockbox itself
23:05:44preglowthen why init in the first place?
23:05:47Bagderboth do the same init I think
23:06:22Bagderthe bootloader needs to init all hw to be able to function properly
23:06:46Bagderrockbox could be made to depend on that, but it currently doesn't
23:07:09preglowsmall problem anyway, yes, just an ifdef should do
23:07:47preglowbut this is extremely cool
23:07:54preglowwe're probably not far away from things starting to function
23:08:27preglowdamn, i'm looking forward to having rockbox boot on my player
23:09:07preglowonly depressing thing is that putting proper emac support in mad is probably going to be a bit hairy
23:09:33preglowwithout having hack the code to pieces, that is
23:10:18preglowthe best thing would be to just have another set of defines in fixed.h, but that won't use the unit optimally
23:11:58 Join webguest00 [0] (
23:12:37 Join webguest19 [0] (
23:12:37 Quit webguest00 (Client Quit)
23:13:36 Join amiconn [0] (
23:13:46 Part webguest19
23:14:20amiconnhi all
23:16:15Bagdergood evening
23:17:05amiconnCassandra: are you there?
23:22:04 Join [IDC]Dragon [0] (
23:22:46[IDC]Dragonhi there
23:22:58amiconnhi J÷rg
23:23:47[IDC]DragonGerd's charging patch looks promising (without looking into it)
23:24:39Bagdergood to see work happen in that area
23:24:47amiconnHmm, I dunno.
23:25:28[IDC]Dragonit's on my list since a long time, but below "threshold"
23:25:48amiconnI had some ideas for the power management too, but going into a different direction
23:26:03[IDC]DragonI was close to start a wiki document today
23:26:15[IDC]Dragonto collect proposals
23:26:32[IDC]Dragonbut no time, and then came Gerd
23:26:40amiconn- Check power much more frequently than now, at least once per second
23:27:12[IDC]Dragonat least in USB mode
23:27:27[IDC]Dragonto detect a spinning disk, if it wasn't us
23:27:30Cassandraamiconn: I'm here, yes.
23:27:31amiconn- Let the power management gather more info about power states, e.g. whether the hd is spinning, whether the backlight is on
23:28:02[IDC]Dragonthat would have been my 2nd stage: try to model the power situation
23:28:09amiconn- The external voltage is also available, so the charging algo could estimate the charging current
23:28:32amiconnCassandra: I know what that "something" in rtc_init() is that resets the alarm flag.
23:28:40[IDC]Dragon1st I would get rid of the startup charging gap, deep discharge, no trickle
23:28:56Bagderyeah, those options don't make much good
23:28:58CassandraOh, right. I hope it made sense to program around it in the way I did.
23:29:02CassandraWhat is it?
23:29:43[IDC]Dragonwe should always start charging, unless we find later it's not necessary
23:29:47amiconnrtc_init() calls rtc_enable_alarm(false); in the end, and this in turn reads the alarm flag, thereby resetting it
23:30:01CassandraOh, right. Of course it does.
23:30:54amiconn[IDC]Dragon: This could be tricky, because charging right away leaves no chance to get the unbiased (by charging current) battery voltage
23:31:20CassandraI couldn't figure out how to do record on power on like we were discussing last night. Feel free to leap in if you can see an obvious way to do it.
23:31:37[IDC]Dragonamiconn, maybe we don't need this
23:32:21amiconn[IDC]Dragon: My idea for the startup is to check voltage first, and enable charging for a short time first (a minute perhaps) only if the voltage is low.
23:32:35[IDC]Dragonwe should calculate the charging current from the DC and battery voltage
23:33:08[IDC]Dragonthen we know how strong we're pulling
23:33:18amiconnYes, but you need the battery voltage without charging current to estimate the battery percentage.
23:33:23Cassandrabtw, speaking of matters power management related, do I misunderstand the code, or could a short idle timeout cause you to lose settings that haven't been flushed to disk (since the idle detection is in firmware and doesn't use the safe shutdown code).
23:33:26[IDC]DragonI'd like to prevent the flat battery problem
23:33:57[IDC]Dragonrather enable charging already in the bootloader
23:34:18amiconn[IDC]Dragon: Yes, that's why I would charge right from the start for a minute or so if the batteries are low, then stop & check voltage.
23:35:09[IDC]Dragonah, ok, but then you've perhaps pulled the cells "above the worst" already
23:35:09amiconnIf battery voltage is high enough, don't enable charging at startup, but immediately calculate battery state, then start charging if needed
23:36:17[IDC]Dragonmeasuring while the HD is spinning may give a good estimation
23:36:25amiconnThe battery voltage with charging current applied doesn't tell you much even when you know the current (by also measuring the input voltage), because you don't know the inner resistance of the cells
23:37:03[IDC]Dragondo you think it greatly varies
23:37:46amiconnYes,definitely. It's highly dependent on charging state. Furthermore it varies across manufacturers, depending on temperature etc.
23:39:16[IDC]Dragontricky subject
23:39:23amiconnIn fact, I also thought about improving the battery state estimation by also calculating the inner resistance (measuring the delta v caused by a certain delta i)
23:40:12[IDC]DragonI'm not too worried about the gauge
23:40:38[IDC]Dragonthat's 2nd, first I'd like to have a reliable charging
23:41:35[IDC]Dragonour charge current is rather low, we don't need too fancy algorithms like if we'd be a high speed charger
23:41:39amiconnIt's not about the gauge, that's a side effect. It's about the charging algo, since, as Gerd pointed out, the usual delta-v algorithm doesn't work well, because of the variable load
23:42:11[IDC]Dragonif in doubt, charge ;-)
23:42:37amiconnOur charging current is certainly higher than what cell manufacturers would allow as permanent charging current
23:43:20[IDC]Dragongo to trickle after a time/voltage limit or if the filtered voltage doesn't raise any more
23:43:50amiconnHow do you calculate the time?
23:44:23amiconnYou'd need the (net) charging current, the battery capacity, and the current charging state of the cells.
23:44:27[IDC]Dragongot me, most simple by capacity and current
23:44:56[IDC]Dragonapart from the start charge, the got me part
23:45:11amiconnIf the cells are about 80% full, and you feed it another 100%, you definitely overcharge
23:45:36[IDC]Dragonthat's where the no-raise detect comes into play
23:46:08[IDC]Dragonor a cutoff voltage
23:46:17amiconnYes, and that's very problematic in the jukebox environment with its highly variable load, as Gerd pointed out
23:46:53amiconnCutoff voltage doesn't help at all, because the chage-end voltage also varies with temperature
23:47:18amiconn(Much more than it varies with charging state, that's the badness with NiCd/NiMH charging)
23:47:48[IDC]Dragonok, so we need a reliable no-raise detector
23:48:09amiconnBelieve me, I once built a charger that only relied on cutoff voltage, and it charged with 1/5 C only
23:48:40amiconnI almost instantly fried 2 sets of almost-new NiCd cells with it....
23:49:00[IDC]Dragonbad cells
23:49:45amiconnNo, these were branded cells
23:50:00[IDC]Dragonactually, I don't understand why 1/10 C should still be allowed for unlimited time
23:50:19[IDC]Dragonwith the capacity rising to 2500 mAh
23:50:20amiconnNo, it shouldn't, it's too much
23:50:34[IDC]Dragonmake it 1/20 C then
23:50:49amiconnAnd for our charging current to be <= 1/10 C, we would need 3500 mAh cells...
23:51:08[IDC]Dragonmy point is, this goes up with capacity, but the AA size is the same
23:51:23[IDC]Dragonit all has to be dissipated as heat
23:51:59[IDC]Dragonand AA cells started at 500 mAh
23:52:37[IDC]Dragon(long time ago, when I was playing with motorized toys ;-)
23:52:46 Join amiconn_ [0] (
23:53:07 Quit amiconn (Nick collision from services.)
23:53:08 Nick amiconn_ is now known as amiconn (
23:53:47amiconn(1) AA cells started even lower than 500 mAh, I have seen 450 mAh cells (still working btw!)
23:53:58amiconn(2) The heat is only part of the problem
23:54:10Cassandrabtw I sort of figured out two ways to implement concurrent sleep timer and wakeup alarm.
23:54:20CassandraThe problem is they both have drawbacks.
23:54:51 Join Trevmar [0] (
23:55:03Cassandra(1) Defer actually setting the rtc until you power off (eg in safe_shutdown)
23:55:17TrevmarJoerg, did you see the pictures I posted on the Ondio forum today?
23:55:35Cassandra(2) Set immediately but have a watch thread that switches it off if we ever get closer to it than a minute.
23:55:52[IDC]Dragonhi Trevor, yes, nice
23:56:04Cassandrasetting the rtc *alarm*, I mean.
23:56:57Trevmarthe hole in front of the mike makes a big difference :)
23:56:58amiconn[IDC]Dragon: When the cells are full, and you still apply charging current, electrolysis takes place, producing hydrogen and oxygen. This has to be absorbed again by the "charging reserve", otherwise the cell looses electrolyte.
23:57:23amiconnIf the current is too high, this doesn't work fast enough.
23:58:09[IDC]Dragonwe could start to trickle when approaching
23:58:14amiconnCassandra: I would prefer option (1) - much more straight forward
23:58:14 Join Digital007 [0] (
23:58:24[IDC]Dragonso, "high" current to empty cells, less while they fill
23:58:45amiconnCassandra: And, yes, you're right, idle poweroff should also use safe_shutdown() (imho)

Previous day | Next day