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

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

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

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

#rockbox log for 2010-10-13

00:05:19 Quit komputes (Quit: I haven't slept for ten days, because that would be too long.)
00:08:18 Quit n1s (Quit: Lämnar)
00:14:44 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
00:17:52pixelmalinuxstb: but what about the sims? Seems the mapping of some buttons it was responsible for doesn't work anymore in at least the D2 sims as reported
00:19:25pixelmamapping of "buttons" - those related to the 3x3 grid mode things
00:22:17 Quit bertrik (Ping timeout: 245 seconds)
00:23:01 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
00:29:10 Quit stripwax (Ping timeout: 240 seconds)
00:35:27 Join stripwax [0] (
00:36:05Torneso are we frozen now or are we still debating? :)
00:38:01 Join casainho [0] (
00:41:39linuxstbTorne: I think we've agreed to freeze, but no-one has said "We are now frozen".
00:44:46 Join JdGord [0] (~jd@
00:45:18JdGordAlexp: pixelma: a noframe option can't be added as simply as the other ones
00:46:04Tornelinuxstb: fair enough
00:46:16JdGordYou could far it with an image though fairly easily
00:48:23 Quit bertrik (Quit: :tiuQ)
00:48:29stripwaxwas the suggestion to freeze on Monday (or was that supposed to be yesterday rather than next Monday?)
00:48:30 Quit Llorean (Quit: Leaving.)
00:49:25 Join aexin [0] (eivindsy@unaffiliated/aevin)
00:49:37 Quit pamaury (Remote host closed the connection)
00:50:25 Quit aexin (Client Quit)
00:52:52 Quit Rob2223 (Quit: Rob2223)
00:54:19linuxstbstripwax: The Monday that's just passed.
00:54:25 Quit casainho (Quit: ChatZilla 0.9.86 [Firefox 3.6.10/20100922073514])
00:57:11 Join Rob2222 [0] (
01:03:53 Join JdGordon| [0] (
01:03:53 Quit JdGordon| (Changing host)
01:03:53 Join JdGordon| [0] (~jonno@rockbox/developer/JdGordon)
01:04:51JdGordon|arg, anyone seen playback freeze recently? the main thread is working fine, playback has just stopped mid track
01:05:16***Saving seen data "./dancer.seen"
01:05:24JdGordon|buffering debug screen says everything is full
01:05:54 Quit JdGord (Quit: Bye)
01:07:34gevaertsI haven't seen that sort of thing recently, no
01:08:14JdGordon|I updated my mini2g for the first time in weeks last night, hopefully its not a recurring thing
01:14:32CIA-7New commit by 03theseven (r28261): iPod Nano 2G USB: Make sure the USB core is properly powered off on boot.
01:15:29chattrhello. on linux kernel 2.6.28-19-generic from ubuntu. sansa fuze v1, sansa firmware v01.02.31A, set to MSC mode. rockbox 3.6 installed today. when I connect the device with the usb cable, I get the constant rebooting mentioned in this channel on Oct 06. any workaround besides booting into the original firmware?
01:15:55gevaertschattr: either install an older bootloader, or a current build
01:16:09CIA-7r28261 build result: All green
01:16:24soapTheSeven, Searching back the las few days I didn't see you mention this latest commit. Should this change create any user-visible improvements?
01:16:25chattrgevaerts: ty. any gotchas with a current build?
01:16:41chattrif you know
01:17:21gevaertschattr: some people (but definitely not everyone) suffer from FS #11608
01:17:39TheSevensoap: it fixes an occasional bug that reduces battery runtime a lot when it occurs
01:17:39gevaertsTry it and see
01:18:33soapahh, some people's USB core was inadvertently staying on post boot?
01:19:10TheSeventhe battery impact was rather huge. the current was 2 to 3 times as high as it should have been
01:20:28 Quit dfkt (Quit: -= SysReset 2.53=- Sic gorgiamus allos subjectatos nunc.)
01:21:01soapdid the USB suck all that power itself, or was there some cascading effect?
01:21:54TheSeveni can't tell what exactly was the power sucker, there might also have been additional system bus load, but most of it was surely eaten by the USB core and clock generation
01:22:11 Quit Xerion (Ping timeout: 252 seconds)
01:22:19TheSeventhis thing eats like 30mA on a device that needs ~15mA total when playing music
01:22:59saratogaFWIW I dislike how strongly rockbox utility pushes people towards release builds, when in fact its fairly rare that theres any serious stability difference between release and current builds
01:24:04soapsaratoga, if one goes much further down that road isn't one basically questioning the whole value of release builds?
01:24:36saratogasoap: IMO we should just say that releases are tested and teh bugs documented, while current builds are not tested but generally have more fixes/features
01:24:44stripwaxsaratoga - fyi FS #11666 in its current incarnation appears to resolve the problems I had running the fastermdct tremor patch on alternative arm devices
01:24:52saratogastripwax: nice!
01:25:24saratogafwiw i wonder if the mdctlib optimizations even compile on non-gcc systems
01:25:27stripwaxsaratoga - not observed any downside yet so going to commit it into rockbox svn in a sec
01:25:30stripwaxsaratoga - i doubt it
01:25:36JdGordon|saratoga: we had a bit of a discussion last night about keeping better track of a stable branch
01:25:54saratogayes i skimmed that
01:28:08 Join Hillshum [0] (4ba5f419@gateway/web/freenode/ip.
01:28:56HillshumTo compile ARM targets on Cygwin, should I use arm-elf-gcc or build the eabi toolchain?
01:29:20CIA-7New commit by 03stripwax (r28262): Rearrange and remove some manual register assignments, make use of pointer address increments for free on arm, and remove macros in favour of explicit ... failed to pull it off, I'll try again
01:30:04stripwaxHillshum - unfortunately the eabi is not prebuilt at
01:30:42gevaertsDid you install all prerequisites?
01:30:53HillshumI think. I should check again
01:31:07CIA-7r28262 build result: All green
01:31:11saratogaunless you're good with cygwin you should probably not use cygwin
01:31:51stripwaxcygwin builds just take forever
01:32:17HillshumI'd rather not get a VM up and running, and I'm not familiar with the other options
01:32:33stripwaxI left my laptop on overnight building gcc (via, and it still wasn't done when I left for work in the morning
01:32:56HillshumWhat are the prerequisites?
01:33:58chattrgevaerts: thanks for the pointer to the flyspray#. I just read the comments. I will wait on trying the current build or an older bootloader, and stick with booting into the original firmware.
01:34:45CIA-7New commit by 03torne (r28263): FS #10910: Manual incorrectly claims that the sleep timer and wake-up alarm are mutually exclusive ...
01:35:16gevaertsHillshum: newer gcc wants libgmp3, libmpfr and texinfo
01:35:32Tornedidn't we fix it to not need those?
01:35:34Torneit doesn't really need them
01:35:45Tornei thought someone added a patch to fix gcc's configure
01:35:56HillshumWould it be possible to integrate wakeup into the alarmclock plugin?
01:36:15 Join perrikwp [0] (
01:36:25CIA-7r28263 build result: All green
01:38:23TorneHillshum: what do you mean?
01:38:32 Part toffe82
01:40:17HillshumTorne: From what I understand, I must decide beforehand if I will allow my Clip+ to power off before the alarm is scheduled to go off and use the plugin if I won't be powering down, or the setting from the System menu if I will. 'Twould be nice if there were only one place to set alarms that didn't care
01:40:56Tornethat's not how it works.. at least it shouldn't be
01:41:15Tornethe RTC interrupt should trigger the alarm anyway
01:41:17 Quit LambdaCalculus37 (Quit: Fwump)
01:41:25Tornethe alarm plugin is only for targets with no RTC, no?
01:42:07HillshumIt's on the Clip+
01:42:25TorneHm, wait, what? it's only built for targets *with* an RTC? that makes no sense
01:42:45JdGordon|no, that cant be right
01:42:49Tornei thought the whole point of the plugin was that it just sat there in the foreground waiting until a given time passed
01:43:10Torneso you didn't need to have an RTC for it to work
01:43:13TorneJdGordon|: it's #if CONFIG_RTC
01:43:28JdGordon|wtf? :p
01:43:45 Join fenugrec [0] (
01:44:51Tornebizarre. apparently i've totally misunderstood what the plugin is for
01:44:57Torneok, so that seems entirely useless then
01:45:34fenugrecHi, running #28212 on a FuzeV1 here, and certain MP3s have warbled audio and playback problems, however I can read those same files on my PC correctly. Anyone else with this problem ? or clues ?
01:45:48HillshumTorne: On targets without RTC, it would be more like an egg timer.
01:46:00fenugrecI was using #26??? before without problems.
01:46:19 Quit stripwax (Quit:
01:46:22fenugrec*24426 rather
01:46:22TorneHillshum: yes, that's what i assumed it *was*
01:47:05Torneokay, er, weird
01:47:15Tornethe rtc drivers define a function to poll if an alarm has happened
01:47:21JdGordon|fenugrec: ALOT has changed since then, the best help you can do is try to go back through those revisions and work out when it started dieing
01:47:22HillshumInstead, you specify a time that it starts playback on. And the wake-up feature requires the thing to be off
01:47:24Torneand the power management thread calls it every second
01:47:35Torneand if it returns true then it.. turns the alarm off. and does nothing.
01:47:41JdGordon|if you can build your own builds it wont take all that long
01:47:42Torneso it specifically detects the alarm going off
01:47:46gevaertsfenugrec: maybe?
01:47:47Torneand then does nothing about it
01:48:07 Quit kazaik (Ping timeout: 245 seconds)
01:48:11HillshumPerhaps it should start playback?
01:48:14fenugrecJdGordon: Arg, I expected that. Is there an online repo of pre-builds ? I totally hosed my rb environment since the switch to "elf-eabi" etc.
01:48:47JdGordon|not that go back that far
01:48:56JdGordon|you can try the 3.6 stable build though
01:48:58TorneHillshum: i assumed it did, but it doesn't look like it.
01:49:11TorneHillshum: so yeah, i don't see any reason why the alarm clock plugin should even exist
01:49:13fenugrecgevaerts: hmm not quite. It's not as severe
01:49:26Tornethe code should just actually *do something* when the alarm goes off while the player is turned on
01:49:32fenugrecuh did the bootloader change a lot since r24* ?
01:50:01TorneHillshum: well, you can get right on that, anyway ;)
01:50:01pixelmaTorne: the alarm thing - isn't (or shouldn't) this still be true for some targets? I thought I remember some hardware limitations (but no details)
01:50:10Tornepixelma: it's possible
01:50:18Tornepixelma: but i checked and none of them turn off as far as I can see
01:50:21HillshumTorne: How much C would I need to know...?
01:50:35Tornevery very old code back in archos-only days did turn it off when the alarm was set
01:50:49pixelmaTorne: then it should still stay in the manuals of affected targets, no`?
01:50:58Tornepixelma: I just checked and there are no targets that do that
01:51:02Torneunless i am missing something
01:51:07JdGordon|pixelma: re the %Vf() bug... your test skin works on the e200 so it could be a greyscale only bug.. does %Vf() work as expected at the start of a viewport at all?
01:51:14Tornethe alarm code is totally platform independant and the rtc code doesn't turn it off
01:51:22JdGordon|that is the last new skin bug so i'd like to close it one way or another
01:51:39Torneit sstill has code in there to prevent you from setting an alarm in the next two minutes, commented saying that it's to stop the alarm from going off during shutdown
01:51:46Tornebut there's no code path that leads to a shutdown
01:52:01Tornethat code is just still there from r5000 or so
01:52:10pixelmaTorne: I'm not sure, I thought the Archoses needed it but my Archos doesn't have an RTC so can't tell for sure
01:52:10HillshumThat's old...
01:52:33pixelmaso s/Archoses/Recorders
01:52:39HillshumTargets need more that just an RTC for this right?
01:52:48TorneHillshum: they need an RTC with an alarm, yes
01:53:03Tornepixelma: i guess maybe the alarm only works as a wakeup source on the recorder?
01:53:13Tornei.e. it doesn't genrate an interrupt or any pollable status value if the player is on
01:53:16Tornethat's possible.
01:53:31Tornebut still i don't see any code that turns it off.
01:53:37Torneand there's certainly no interaction between the alarm and sleep timer
01:53:39fenugrecJdGordon: should I load a more recent bootloader than 24* ?
01:53:53pixelmaTorne: I think it was that it immediately needed to turn off but I'm really really unsure
01:54:12Tornepixelma: well, i'm al;most certain that it *doesn't* turn it off. maybe it does need to ;)
01:54:16pixelmaJdGordon|: ok, I'll try in a minute
01:54:19Tornebut if so then it looks like the code is broken
01:54:52JdGordon|fenugrec: I dont know the fuzes very well... not sure. I dont tihnk it makes any difference though
01:54:57JdGordon|as long as it boots at all
01:55:08fenugrecok thanks, I'll do some more testing this evenin'
01:55:13Tornepixelma: it doesn't turn it off on any of the targets i have, anyway ;)
01:55:19pixelmaIIRC only the V2/FM Recorders could wake up and some could be modded
01:55:22fenugrecthanks, I may be back
01:55:31 Part fenugrec ("Leaving")
01:57:51HillshumOkay. Here's my Cygwin compilation code
01:58:28gevaertsAh yes, you also need libiconv
01:59:12pixelmaJdGordon|: doesn't work with r28231, should I expect something from the later changes?
01:59:28JdGordon|pixelma: what doesnt work?
01:59:44*Hillshum meant to try installing that....
02:02:08pixelmathe test.wps - should change foreground greyshade with hold status for the current playing time. The latter shows up but doesn't change shade and the "TTT" etc. inside the conditionals don't show up at all
02:02:25Torneis it too horrible to call bookmark_autobookmark() from powermgmt.c ? :)
02:02:40TheSevenerm, WHAT?
02:02:44JdGordon|pixelma: right, can you check if %Vf works as it should outside of condiaiontlas?
02:03:09Tornepoweroff from the sleep timer doesn't save bookmarks at the moment; it's trivial to fix for the case where you actually power off by just removing the call to audio_stop() in there (shutdown stops audio anyway)
02:03:19Tornebut for the case where you're plugged in and can't power off.. not sure what else to do
02:04:06pixelmaJdGordon|: it does with my usual WPS using conditional viewports
02:04:40JdGordon|ok, I wonder why grey is being wierd then
02:06:56 Quit leavittx_ (Ping timeout: 264 seconds)
02:07:29pixelmaat least stack usage doesn't jump to 90% anymore when going to the WPS and stays at 50% on the M5
02:10:23 Join fdinel [0] (
02:12:24pixelmaI'll update and see if something changed
02:12:31 Quit Judas_PhD (Quit: This is a quitting message)
02:14:45 Join Xerion [0] (
02:14:59pixelmaI see it working on my colour target, the c200 which I just updated. Back when I reported it, the thing didn't work in an X5 sim
02:16:22JdGordon|it still wont work with static text, but there is a simple fix for that... It needs a tag which always causes the line to change but not draw anything which could be added ismply
02:16:33JdGordon|but it not working at all for grey is wierd
02:17:22pixelmaI'm currently compiling a new revision, don't know what happened in the last 30
02:17:47JdGordon|nothing which would change that tag
02:19:23 Quit Xerion (Read error: Connection reset by peer)
02:20:05 Quit Depressicaa (Ping timeout: 252 seconds)
02:20:09pixelmaI really took the WPS from my M5 and copied it to the c200 and just replaced the grey values with colour ones so there shouldn't be a wps code problem. Hmm, by the way the text inside the conditional doesn't show up either on colour, not sure if it should
02:20:33 Join Nausicaa [0] (
02:20:38pixelmajust the playback time which changes colour on switching hold
02:20:38 Join Thanatos [0] (
02:20:56JdGordon|yeah, the text after Vf not showing is odd, I'm not sure if it is expected or not
02:21:07JdGordon|from a user POV it obviosuly isnt expected
02:22:12 Quit Zambezi (Quit: 2weeks idletime.)
02:24:54HillshumI got compiling the eabi toolchain on Cygwin.
02:25:18JdGordon|same error S_a_i_n_t was getting
02:26:04pixelmaso, if I have battery level displayed in % and want it to turn red for the last 10% - would having the %Vf inside the battery level conditional work if the only text the will be updated in such a case is battery level in % - at the same number it would change colour or maybe if it decreases on more percent?
02:27:40JdGordon|it shuold
02:27:52pixelmawhen it changes "state", if and when it changes colour is the actual question
02:28:52pixelmaI think I could simplify my WPS that why, will play around with it a bit the next days
02:29:28JdGordon|%bl%%%if(%bl, <, 10)<%Vf(ff0000)|%Vf(aaaaaa)>
02:29:32JdGordon|I would expect that to work
02:30:03S_a_i_n_twhere's the eabi toolchain *supposed* to live?
02:30:05JdGordon|%bl%% could be on either end
02:30:17pixelmaok, just wanted to ask about that
02:30:32JdGordon|as long as its on the same line it should work
02:30:36S_a_i_n_tmy PC died last night from loss of power (powercut) and I've no idea if it finished building or not.
02:30:49JdGordon|(should here being "I would hope it would")
02:31:25pixelmait needs to be on the same line?
02:31:37JdGordon|well no
02:31:53 Join webguest83 [0] (
02:32:01JdGordon|only lines *after* the Vf change *and* have dynamic tags will change colour
02:32:49JdGordon|so unless a special tag is added (or some rather messy code to check colour changes) anything more complex than a single line is better with seperate viewports
02:35:38pixelmareally no change on my M5, don't know how high the chances are that it is a remote target problem maybe (though grey remote too). If you want to, you could check it on your Mini I guess
02:35:39 Join krabador [0] (
02:36:31JdGordon|i tried on the mini sim last night and it looked like Vf wasnt working at all...
02:40:21S_a_i_n_thahahaha....that's really funny with the alarm plugin.
02:40:56S_a_i_n_tLlorean tried ripping a shred off me yesterday for suggesting that wasn't intended function.
02:41:44 Quit timccc (Quit: Leaving.)
02:41:46S_a_i_n_tthe discussion about the point of the alarm came up yesterday, and if it should/shouldn't wake the device.
02:42:33pixelmathen it's not about the plugin
02:42:43 Quit saratoga (Quit: Page closed)
02:45:28 Quit webguest83 (Quit: CGI:IRC)
02:46:08*S_a_i_n_t isn't quite sure what to say there...
02:46:44S_a_i_n_tseems Torne and Hillshum had the same discussion, only it went a lot better.
02:47:40pixelmaJdGordon|: %Vf only seems to work if it is on the same line as the %V
02:48:23JdGordon|on grey? or all targets?
02:48:37pixelmaon grey it seems
02:49:18JdGordon|I wonder if there is a incorrect #if somewhere then
02:51:34pixelmaI played around with the test.wps - just removing the conditional statement only leaving %Vf(2) on an own line first and it didn't work. I wondered why my full WPS works then and then appended the %Vf to the viewport line and it started working
02:52:08pixelmasorry for weird English, I should go get some sleep
02:58:37 Join anewuser [0] (kvirc@unaffiliated/anewuser)
03:05:14 Quit Hillshum (Quit: Page closed)
03:05:18***Saving seen data "./dancer.seen"
03:27:17S_a_i_n_tJdGordon: With your 180x800 backdrop. did you want to use a larger font than the cabbie port for android already uses?
03:27:32S_a_i_n_tI mean, will the current "titlebar backdrop" be sufficient?
03:27:55S_a_i_n_tOr do I need to make it larger>
03:29:33S_a_i_n_terrr...yeah, oops.
03:29:41JdGordon|I want it the same as the wps, so either really
03:29:57JdGordon|I dont actually mind the smaller title text
03:30:05JdGordon|need to see what it looks like on target thoguh
03:33:37 Join timccc [0] (~tim@
03:43:30S_a_i_n_tJdGordon: far too easy ;)
03:43:45JdGordon|I know :)
03:43:59JdGordon|please please please can you fix the wps popup though?
03:44:13JdGordon|without the stuff at the bottom
03:44:23S_a_i_n_tUm...yes, what's needed to "fix" them?
03:44:38S_a_i_n_tOh, ok...I'll have a look.
03:44:42JdGordon|a area for a mein button, and a hotkey button
03:45:30S_a_i_n_tthe .sbs background should be what you wanted...I'd probably go so far as using it for the .wps too and adding the titlebar in with a viewport so all the fonts match throughout the screens.
03:45:48S_a_i_n_tI never really liked the idea of the title being part of the backdrop.
03:46:41JdGordon|it should be translated
03:46:47JdGordon|importing that font would be good
03:47:22S_a_i_n_tI *try* to add translation anf rtl support wherever possible.
03:47:38JdGordon|mind you, specifying fonts in the skins is bad for translation support
03:48:17S_a_i_n_tIt is, yes...a serious workaround is needed for the fonts. in theory they all should have the same charsets.
03:48:30S_a_i_n_tthat would be nice, but a LOT of work.
03:48:48S_a_i_n_t1000s of chars created by hand, uuuurgh.
03:48:54JdGordon|yeah, not gonna happen
03:49:11JdGordon|still this way is worth that problem anyway
03:49:43S_a_i_n_tIt may be possible now with %if (perhaps not) to use unifont for rtl cases?
03:50:14JdGordon|well, the viewport font could be changed like the colours
03:50:17S_a_i_n_tI always wanted a way to select fonts conditionally, but, biggie.
03:50:52S_a_i_n_tthen it could fallback to unifont for rtl, which is the only font that supports rtl.
03:51:02JdGordon|no it isnt
03:51:20S_a_i_n_toh...hmm, what else does? that's good to know.
03:51:48JdGordon|"> hebrew and arabic are the obvious rtl langs
03:52:03JdGordon|hebrew is pretty well covered
03:52:22JdGordon|arabic barely
03:53:11JdGordon|I'm guessing the ~30% hebrew coverage is just letters and not the optional pronounciation dots
03:53:17S_a_i_n_twow. I'd never seen that link before, my pissing around lead me to conclude unifont was the only font that supported rtl.
03:53:58JdGordon|a font doesnt support rtl or not... it supports glyphs
03:54:03JdGordon|which some languages would draw backwards
03:54:18S_a_i_n_twell, yeah. bad wording on my part.
03:55:59S_a_i_n_tand I guess what I meant by "supports" was "100% of possible glyphs"...I'm aware that there are fonts with partiial coverage for rtl langs.
03:56:37JdGordon|yeah, but like I said, hebrew is perfectly usable with the 30% covergae it has
03:58:24 Quit DerPapst (Quit: Leaving.)
04:00:51S_a_i_n_tJdGordon: With the popup, does the whole image stay the same size? (i'm guessing so) and I need to make room inthe space there is? or can the image size be increased? (probably mess up your positioning)
04:01:09JdGordon|up the image size I tihnk
04:01:27JdGordon|also, if oyu can be bothered, the browse image seems bigger than the other two which is a bit odd
04:01:45JdGordon|I can fix the positioning easily
04:01:45S_a_i_n_tyeah, I was just looking at that.
04:01:51S_a_i_n_tok, cool.
04:04:22JdGordon|can you do both screen sizes in svn please? 320x480 and 480x800
04:05:05 Quit perrikwp (Quit: CGI:IRC (EOF))
04:05:58 Quit pixelma (Disconnected by services)
04:06:00 Quit amiconn (Disconnected by services)
04:06:00 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:06:00 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:06:02 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:06:15S_a_i_n_tYeah, I can...but it'll take me a little longer than the backdrop fix did ;)
04:06:20 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:06:21S_a_i_n_tA little more complicated ;)
04:06:28 Join perrikwp [0] (
04:10:25 Quit TheSeven (Ping timeout: 240 seconds)
04:12:12 Quit rvvs89 (Ping timeout: 265 seconds)
04:12:30 Join rvvs89 [0] (
04:14:37 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
04:16:31 Join Judas_PhD [0] (
04:48:28 Join dys` [0] (
04:49:21 Join Barahir [0] (
04:49:53 Quit MethoS- (Remote host closed the connection)
04:50:49 Quit dys (Ping timeout: 276 seconds)
04:52:48 Quit Barahir_ (Ping timeout: 265 seconds)
04:55:20 Quit Strife89 (Quit: Sleep cometh near.)
05:03:17 Quit froggyman (Quit: Bye)
05:04:47 Join froggyman [0] (~seth@unaffiliated/froggyman)
05:05:07 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
05:05:20***Saving seen data "./dancer.seen"
05:06:39 Quit Farthen (Ping timeout: 252 seconds)
05:14:18 Join soap_ [0] (~soap@rockbox/staff/soap)
05:15:21 Quit soap (Disconnected by services)
05:15:24 Nick soap_ is now known as soap (~soap@rockbox/staff/soap)
05:20:53JdGordon|eeek! PANIC "Write 710555265 after data"
05:33:47 Quit ps-auxw (Ping timeout: 240 seconds)
05:44:28 Quit Horschti (Quit: Verlassend)
05:45:13 Join ps-auxw [0] (
05:47:29JdGordon|oh crap, got that again... /me wonders who broke it
05:48:27JdGordon|sweet, simple repro.. start some music. pause it then start another track in the file broswer
05:48:33JdGordon|can anyone else repro this?
05:54:07JdGordon|never mind. Apparently .playlist_control got toally screwed up!
05:54:08JdGordon|/.rockbox/.playlist_controlAA Start cluster beyond limit (18446744073196864544 > 1978987). Truncating file.
05:54:11JdGordon|/.rockbox/.playlist_controlAA File size is 37267 bytes, cluster chain length is 0 bytes.
06:08:18 Quit fdinel (Quit: Miranda IM! Smaller, Faster, Easier.
06:25:06 Quit S_a_i_n_t (Disconnected by services)
06:25:09 Join S_a_i_n_t [0] (S_a_i_n_t@
06:38:27JdGordon|S_a_i_n_t: thinking about skinning the list items... how does this sound...
06:38:53 Quit edboyer93 ()
06:39:06JdGordon|there would be a tag to say which viewport label to use for unselected list items, and to say which to use for the selection
06:40:52JdGordon|you would then setup the viewports for both of them, but instead of the viewport x,y being screen relative they would be relative to the main list item selection viewport
06:42:24JdGordon|e.g %list_item(x,y,width,height, <selection vp label>, <normal item vp label>) <- x,y there is screen relative (so as expected)
06:42:51S_a_i_n_tIt sounds....really complicated, and I can't even begin to imagine it in code form...but, that's not my job so "OK" ;)
06:43:12JdGordon|then you'de setup viewports like %Vl(selection, 0,0,10,10,0) for an icon (which draws at 0,0 in the selection)
06:43:33JdGordon|%Vl(selection, 0,10,-,10,0) would be below it... etc
06:44:12 Quit S_a_i_n_t (Disconnected by services)
06:44:14 Join S_a_i_n_t [0] (S_a_i_n_t@
06:44:32JdGordon|you're not kidding complicated :) this is the problem, I cant tihnk of a simple way to do it
06:45:44JdGordon|but doing this would get rid of I tihnk every complaint there is with the list
06:45:56JdGordon|vertical spacing, nicer selection bar, bmp selection bar
06:46:27JdGordon|2 line selector lists are a bit of a WTF with this though
06:48:42 Quit S_a_i_n_t (Ping timeout: 265 seconds)
06:48:55 Quit Chronon (Quit: Fnord!)
06:49:18JdGordon|I guess to start with multiline selectors would just have to use the selection vp more than once
06:52:35 Quit Judas_PhD (Quit: This is a quitting message)
06:53:34 Join S_a_i_n_t [0] (S_a_i_n_t@
06:54:41S_a_i_n_tSo, um...yeah
06:54:48S_a_i_n_twhy does it need two VPs?
06:55:10S_a_i_n_tdifferent fonts?
06:55:10S_a_i_n_t(sorry...I fell off the infonetz)
06:55:51S_a_i_n_tif you could have a look at transparency for the line selector...I'll love you long time, or not, whatever gets the most incentive.
06:58:32 Join hebz0rl [0] (
07:00:12JdGordon|why does what need two vps?
07:00:23S_a_i_n_twhat you were just talking about
07:00:25S_a_i_n_tthe lists
07:01:28JdGordon|you mean why did my example show two? or why i said a different vp label for the selected and not selected items?
07:01:39S_a_i_n_tthe latter
07:03:10JdGordon|just for flexibility
07:03:13S_a_i_n_tis gcc-core.4.4.4 from the same one as the script pulls?
07:03:35S_a_i_n_tI guess it adds for cool effects with ease too
07:03:35JdGordon|otherwise it would need a tag to check if we are selected or not
07:03:43S_a_i_n_tlike diffent font/colour for selection
07:04:08S_a_i_n_tthe tag to check selection could be useful though, as it could be used elsewhere though
07:04:47S_a_i_n_t(seems kinda lame, but to display a bitmap or notification if selection is happening in the lists)
07:05:19JdGordon|I dont see how that could work
07:05:21***Saving seen data "./dancer.seen"
07:06:12S_a_i_n_tlike "disk acces" except for "are we scrolling/selecting in the lists" or am I imagining what you said about a tag to check selection too literally?
07:08:23JdGordon|it would need to be more "is the viewport being drawn now the selection" more than "is there a selection"
07:08:41JdGordon|which is harder to make work than just seperate viewports
07:08:45S_a_i_n_tAha, right.
07:09:09 Quit panni_ (Quit: ( :: NoNameScript 3.81 :: ))
07:09:10JdGordon|scrolling would be totally fucked though I think
07:09:20S_a_i_n_tfair enough. seperate viewports does make it somewhat easier to configure a selected and non-selected look for list items
07:09:35 Join Farthen [0] (
07:09:47 Quit anewuser ()
07:09:57S_a_i_n_tscrolling would possibly be going off at different times, yes, which could look mighty weird.
07:10:13JdGordon|the way im talking about you couldnt use scrolling at all
07:10:33S_a_i_n_tmajor drawback.
07:11:11S_a_i_n_tthe idea of a selected VP/non-selected VP seems really appealing, if you can find a way to make scrolling work.
07:11:27S_a_i_n_tBut I'm sure completely re-doing scrolling isn't high on your list.
07:12:18JdGordon|nor is making the lcd/viewport drivers suppport child viewports which this really needs
07:18:03 Quit JesusMcCloud (Ping timeout: 245 seconds)
07:26:28 Join JesusMcCloud [0] (~JesusMcCl@
07:26:49 Quit mc2739 (Ping timeout: 276 seconds)
07:28:31 Join mc2739 [0] (~mc2739@
07:28:31 Quit mc2739 (Changing host)
07:28:31 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
07:44:31JdGordon|another reason for 2 viewports is the selected item could be a bigger rectangle than the non selected
07:50:49 Quit hebz0rl (Quit: Leaving)
07:51:30 Nick dys` is now known as dys (
08:01:54 Join Judas_PhD [0] (
08:06:12 Quit Judas_PhD (Client Quit)
08:09:33 Join fyrestorm [0] (
08:09:35 Quit tchan (Ping timeout: 276 seconds)
08:12:48 Quit elinenbe (Ping timeout: 265 seconds)
08:21:59 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
08:22:06 Quit JdGordon| (Quit: leaving)
08:26:06 Join JdGord [0] (~jd@
08:31:07 Join wodz [0] (
08:32:00 Quit Nausicaa (Disconnected by services)
08:32:26 Join Depressicaa [0] (
08:34:17 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
08:34:24 Join Wyatt [0] (
08:35:11 Part Depressicaa
08:35:20 Quit Thanatos (Quit: Leaving)
08:44:52 Quit JdGord (Quit: Bye)
08:45:10 Part Wyatt ("Konversation terminated!")
08:51:30 Part Zagor
08:52:02 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
08:52:11CIA-7New commit by 03bertrik (r28264): Update of Czech language - FS #11674 by Marek Salaba
08:53:01 Join leavittx_ [0] (~lev@
08:53:53CIA-7r28264 build result: All green
08:56:40 Join kevku [0] (
08:58:55 Quit bertrik (Quit: :tiuQ)
09:00:34 Quit perrikwp (Quit: CGI:IRC (Ping timeout))
09:04:28 Quit leavittx_ (Ping timeout: 264 seconds)
09:05:22***Saving seen data "./dancer.seen"
09:08:11*S_a_i_n_t sees another report of undefined instruction whilst skipping tracks on Nano2g:
09:08:39S_a_i_n_tThis is now myself, G4)blivion, and this guy experiencing the issue.
09:09:18S_a_i_n_tI seem to be able to reproduce it the easiest, I can make it happen whenever I want just by skipping tracks too quickly either forward or back.
09:09:18 Quit wodz (Ping timeout: 245 seconds)
09:09:42 Join ender` [0] (
09:10:31 Part Zagor
09:10:45S_a_i_n_tG4Oblivion reported that resetting his config fixed the issue for him, which I haven't looked into yet, but, there shouldn't be ay config value that makes this behaviour more/less just shouldn't happen.
09:11:09 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
09:15:57 Join elinenbe [0] (
09:21:23 Quit linuxstb (Ping timeout: 245 seconds)
09:24:09 Join LinusN [0] (~linus@rockbox/developer/LinusN)
09:24:19 Quit krabador (Ping timeout: 240 seconds)
09:32:01 Join S00row [0] (
09:34:05 Join linuxstb [0] (
09:34:10 Quit linuxstb (Changing host)
09:34:10 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
09:34:24*S00row tips his hat
09:35:53S00rowI'm very new to rockbox, attracted by and the Rockbox as an app andoid port. I'm compiled, installed and running and i'd like to track development of the project.
09:35:59 Join Rob2223 [0] (
09:36:08ZagorS00row: welcome!
09:36:31S00rowPreviously i've done this using rss + git activity lists,
09:37:19Zagoryes, the commit and flyspray mails are a good way to see what's happening. and being here. is close to what i want
09:37:55Zagorwe post all commit on twitter if you prefer that :)
09:39:00 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
09:39:25 Quit Rob2222 (Ping timeout: 240 seconds)
09:39:39Zagorwe don't currently have a commit rss feed
09:41:44 Quit elinenbe (Ping timeout: 265 seconds)
09:42:45 Join swilde [0] (
09:43:25 Join elinenbe [0] (
09:44:17S00rowfeature request ;)
09:45:00Zagoryes, it makes sense to have one. I don't have time to look into it right now though.
09:48:16 Quit linuxstb (Ping timeout: 252 seconds)
09:48:48S_a_i_n_tI thought there was an RSS feed for commits?
09:49:00S_a_i_n_tOr, is that just for 'major changes'?
09:49:38S_a_i_n_tA recent forum post seems to imply rather heavily that there's one of those two at least. I forget which.
09:51:15*S00row needs to learn how to type
09:51:18S_a_i_n_taha, yeah. that's the one.
09:51:30rashercia has one for commits
09:51:30S00rowIt isn't responding at the moment
09:52:00rasherWorks for me
09:52:15 Quit S00row (Read error: Connection reset by peer)
09:52:42pixelmawasn't there also something linked to gmane?
09:54:46 Join S00row [0] (
09:55:16rasherAh yes probably one for the -cvs list seems to have what i'm looking for
09:57:18 Join wodz [0] (
09:57:37wodzpixelma: just nagging :-)
09:57:48ZagorI'll add a link to the front page
09:59:03 Quit sasquatch (Quit: WeeChat 0.3.2)
09:59:09CIA-7New commit by 03zagor (r28265): Added link to rss commit feed
09:59:30 Join sasquatch [0] (
09:59:40 Quit Topy (Read error: Connection reset by peer)
09:59:54 Join Topy [0] (
10:01:12AlexPJdGordon: Yeah, I'm not bothered really about the noframe nobar whatever, I was just trying to understnad them to describe them :)
10:03:56 Quit S_a_i_n_t (Disconnected by services)
10:03:58 Join [Saint] [0] (S_a_i_n_t@
10:07:27 Part LinusN
10:08:13 Join LinusN [0] (~linus@rockbox/developer/LinusN)
10:09:36AlexPZagor: Are the sponsors logos missing, or do we no longer have any and the "Sponsors:" is superfluous?
10:09:40*AlexP guesses the former
10:10:30ZagorI see them
10:10:44AlexPhmmm, not here (chrome on windows)
10:10:48AlexPoh, maybe adblock
10:10:50AlexPone mo :)
10:11:15pixelmayes, adblock (at least here)
10:11:16AlexPZagor: yeah, sorry - adblock :)
10:13:50 Quit pSiKO (Read error: Connection reset by peer)
10:24:26 Quit kevku (Quit: KVIrc 4.0.2 Insomnia
10:24:33 Join kugel [0] (~kugel@rockbox/developer/kugel)
10:29:37 Join pSiKO [0] (~tbox@
10:31:43 Join noamsml_ [0] (
10:34:31 Join DerPapst [0] (~Alexander@
10:34:39 Quit [Saint] (Ping timeout: 240 seconds)
10:35:27 Quit noamsml (Ping timeout: 264 seconds)
10:35:44 Join S_a_i_n_t [0] (S_a_i_n_t@
10:39:45 Quit S_a_i_n_t (Ping timeout: 240 seconds)
10:41:07 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
10:41:54 Join Judas_PhD [0] (
10:45:47 Join petur [0] (d408b802@rockbox/developer/petur)
10:47:39 Quit wodz (Quit: Leaving)
10:56:21 Join xsloader [0] (
10:56:23 Quit linuxstb (Ping timeout: 245 seconds)
10:57:18kugelthe speedup from asm is *huge* for all codecs except applelossless and mp3
10:57:39JdGordonpixelma: OK, I tihnk I see what the problem with Vf is... it isnt a skin bug at all (well yes and no).. the 2bit lcd drivers store the colours when the viewport is set (i.e when the skin sees %V() ), chaning it doesnt take effect till the next %V() call...
10:57:46*kugel wonders how to make a nice chart from it
10:57:57JdGordonso not entirely sure how to fix it cleanly
10:59:16kugelJdGordon: yea, I noticed that too with splashes once. you need lcd_set_viewport or lcd_set_{fore,back}ground everytimes the colors change. it will not apply automatically just from changing the fields in the structs
10:59:48JdGordonits odd, and would be nice to be documented, but I think I can see why its doing that
10:59:54JdGordon16bit is fine, 2bit only
11:00:21JdGordonfg_pattern = 0x55 * (~current_vp->fg_pattern & 3); <- in lcd_set_viewport()
11:00:26JdGordondoing that every time would suck
11:02:01JdGordonso probably the "fix" is to constantly watch the viewport colours in the skin render and reset the viewport if they change
11:02:30 Part Zagor
11:02:44 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
11:04:32kugelwe should definitely store test_codec results as xml!
11:04:43 Quit sasquatch (Ping timeout: 245 seconds)
11:04:56JdGordondoes the asm benefits help with the htc bug?
11:05:24kugelbut not for the majority of people since mp3 isn't faster
11:05:26***Saving seen data "./dancer.seen"
11:16:42 Join Zambezi [0] (Zulu@
11:17:19 Quit Zambezi (Changing host)
11:17:19 Join Zambezi [0] (Zulu@unaffiliated/zambezi)
11:18:35Stummii can't build rockbox after applying the first patch:
11:18:50pixelmaok, maybe close it then as a theme bug. Not sure if the other should be reported as "bug" or not, as a non-coder I just find it weird if things behave differently and am afraid it will be forgotten about. Would be nice if I could do the same rework of the WPS on greyscale as well (it's the same basic design and idea behind, on greyscale a bit simpler as the usable different shades are limited)
11:18:55pixelmaJdGordon: ^
11:19:14JdGordonna, the fix is 3 lines so I'll do it anyway
11:19:27JdGordonIt also partially fixes static text so the tag actually works how you'd expect
11:20:01pixelmaStummi: are you applying both patches?
11:20:20Stummipixelma, no, just the first. I will try it with both then
11:20:44Stummi(it should build with the first patch only)
11:20:51 Join robin0800 [0] (
11:21:06pixelmayou need the enable and the USB one, the first enables something that isn't there if you don't apply the second if I understand correctly
11:22:19amiconnTorne, pixelma: The reason for the alarm code being suppoed to shut down the player is to make sure the device cannot be locked up in a state where it can't be shut down at all
11:22:52amiconnThis would happen on archos fm/v2 recorder (and recorder v1 with alarm mod) if the alarm fires while the device is on
11:23:54amiconnIf that doesn't work anymore it must have gotten lost at some point. This is kinda dangerous...
11:26:05JdGordonpixelma: AH, and I figured out why TTT and GGG arnt being displayed... the Vf and Vb are marked to ignore the \n (so when they are on the same line as %V the text isnt pushed down one line)... because of that the next line is overwriting the text... not sure how/if to fix that
11:27:15JdGordonmaybe that can be fixed though...
11:28:45JdGordonhmm nope.. ok, known limitation
11:31:17 Quit robin0800 (Remote host closed the connection)
11:31:39 Join linuxstb [0] (~linuxstb@
11:31:39 Quit linuxstb (Changing host)
11:31:39 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
11:31:53amiconnJdGordon: Watching would need more cpu than just doing that assignment. It's just a negation, masking, and multiplication with a constant
11:32:16JdGordononce per line in the skins
11:33:08amiconnLine as in line of text?
11:34:02amiconnThat's nothing then
11:34:08JdGordonor do that calc for every pixel? isnt that the same complaint as the background?
11:34:40amiconnIt's not.
11:35:30amiconnDoing a single if() per pixel is far far worse than doing this 3-to-5 machine instructions thing perhaps a dozen times for the whole screen
11:35:55 Quit linuxstb (Ping timeout: 240 seconds)
11:38:43JdGordonpixelma: ok, so here is the choice... currently any text on the same line as Vf/b will never be displayed. I can change it to either *always* add the new line (which will mess up people viewports) or only drop a line when it is with text... so %?mh<Vf(0)> will not add a new line but adding any text (or text tags) will add a new line
11:38:53JdGordonso it is slightly less consistant
11:39:03*JdGordon isnt sure which way to go here
11:39:28Stummipixelma, it was my failure. Just the bootloader seems fails to build after patching. The normal rockbox-build works
11:43:32 Quit kugel (Ping timeout: 265 seconds)
11:45:00 Join S_a_i_n_t [0] (S_a_i_n_t@
11:46:04CIA-7New commit by 03jdgordon (r28266): Fix FS #11593 - %Vf() and %Vb() should be able to be used in conditionals. Remember that these tags cause a new line to get drawn so text/tags on the ...
11:46:11pixelmaJdGordon: about the greyscale thing - if it is possible I'd prefer if the colour and greyscale drivers could handle it the same. I think the more uniform they are/behave the easier it is to deal with them later on. And about the newline, always adding one and messing with existing WPSs is probably the worst of the options. If it is documented somewhere I even wouldn't mind if it stays the way they are
11:46:48JdGordonthey are
11:47:12S_a_i_n_twhat *is* with that newline thing?
11:47:23S_a_i_n_tit caused me a bit of grief before I knew about it.
11:47:39JdGordonsee the example wps in that bug..
11:47:43CIA-7r28266 build result: All green
11:48:35JdGordonpixelma: screen corruption bug... can it be closed?
11:48:46S_a_i_n_twow...a tag other that %Vf or %Vb in a viewport def looks *weird*
11:48:59S_a_i_n_tbut, I'm used to still tying the fg/bg colour in with the VP
11:49:17S_a_i_n_teven though I don't have to, I keep them on the same line out of habit
11:49:41JdGordonyou should do it like that, unless you want them to change
11:50:13Torneamiconn: i can't see any sign that it does that, no, so either it's been lost or i'm blind :)
11:50:40pixelmaJdGordon: I'll test in a bit and close it when I don't see corruption anymore after switching some themes back and forth on my Ondio
11:51:39pixelmamy theme (when it worked) didn't show screen corruption but was missing some images sometimes
11:52:11*JdGordon sticks fingers in his ear... skin code is bug free now!
11:53:09S_a_i_n_t*known* bug free ;)
11:53:24S_a_i_n_twe'll call the new ones "features" from now ;)
11:53:30pixelmaI guess nothing is ever fully bug free
11:53:49JdGordonS_a_i_n_t: you're learning! :)
11:54:10*pixelma remembers when dionoea claimed solitaire being bug free when you could still make the 2 of clubs magically disappear
11:54:24S_a_i_n_t"Reason for closing: I meant to do that!"
11:56:25 Join hebz0rl [0] (
11:57:54pixelmaanyone else experiencing that you need to press a button twice to take effect if things start scrolling on the screen? Seems to happen regularly on my M5 now when wanting to call the menu from the file browser (filter first keypress is enabled but it happens even if the backlight is still on)
11:58:13S_a_i_n_tNot here
11:58:31pixelmadidn't see it on my c200 either
11:58:41S_a_i_n_tanother grey only thing?
11:59:36pixelmacan't imagine it has something to do with the display, more with buttons and how their status are tracked or so. That's just a wild guess though
11:59:44S_a_i_n_thave you tried it without the "filter first keypress" setting? If it's on?
11:59:50*S_a_i_n_t reads, it is.
12:02:33pixelmanot that the M5/X5 is great wrt button handling, I'll watch it a bit more, maybe I was just unlucky yesterday
12:03:07amiconnTorne: Problem here is that the rtc's alarm output is just doing the same as the ON button. There is no interrupt to the host cpu
12:03:34amiconnAnd the rtc only resets the alarm output after acknowledging - which won't happen if it fires while the device is on
12:05:06*amiconn thinks that could be fixed without forcing the device to be powered off after setting the alarm
12:05:24amiconnUnfortunately I don't have such a device
12:11:10Torneright, i guessed
12:18:31 Part Zagor
12:18:45 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
12:35:59 Join dfkt [0] (dfkt@unaffiliated/dfkt)
12:38:12 Quit bug2000 (Ping timeout: 276 seconds)
12:44:16 Quit xsloader (Remote host closed the connection)
12:45:21amiconnI think it should be possible to check alarm status in the shutdown function. If it fired, reset it.
12:45:36amiconnThis would disable hardware poweroff though, in case rockbox hangs
12:46:26amiconn(only way to power down is to pull the batteries then)
12:59:24Tornewell, i am indifferent to how it should work; i was just trying to correct the manual
12:59:39Tornesince someone complained ages ago that the manual said they couldn't use sleep + alarm together
12:59:42Torneand i tried it and it worked
13:00:12Torneso either the archos shuts the player off when you set the alarm and i'm just not seeing it, in which case the manual should probably mention that but only for archos
13:00:32Torneor it doesn't in which case maybe it needs fixing, and maybe the manual should mention it :)
13:00:56Tornebut as far as i can see the manual now matches the code
13:05:29***Saving seen data "./dancer.seen" seting the alarm (alarm plugin, not the time/date settings alarm thingo?) supposed to shut down the player?
13:11:31S_a_i_n_ts/player/DAP/ (avoid confusion with the archos player)
13:11:51JdGordonyes, i tihnk so
13:12:24S_a_i_n_twow...that's been busted for quite some time then.
13:12:40S_a_i_n_tif it is indeed so.
13:17:39 Join teru [0] (
13:17:52 Join anewuser [0] (kvirc@unaffiliated/anewuser)
13:23:31 Join xsloader [0] (
13:25:09 Quit user890104 ()
13:26:49 Join kugel [0] (~kugel@rockbox/developer/kugel)
13:28:28JdGordonkugel: looking at FS #11575, is there any reason we cant have rockbox store a magic string (like "<RaaARoot>") instead of /usr/local/share in the config?
13:28:54JdGordonapart from shortening the string/path stored, it would let you rebuild with a different path
13:29:13JdGordonthen file search order would still be the same
13:29:43 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
13:30:03JdGordonwhere is the code that does the searching?
13:38:34TorneS_a_i_n_t: i checked and it hasn't done that obviously since about r5800
13:38:48TorneS_a_i_n_t: so i have a hard time believing that it's accidental
13:39:01Tornei.e. it was only true in archos-only days
13:42:31 Quit Topy (Read error: Connection reset by peer)
13:45:24JdGordonis "/<>/" always going to be an invalid filename? or is that only FAT?
13:46:55Torneonly FAT
13:47:05TorneWell, okay, fat and unix
13:47:11Torneyou can't have a / in a filename on unix
13:47:32 Quit kugel (Ping timeout: 260 seconds)
13:47:50TorneOh, right
13:47:56JdGordonI'm after a magic string which would never be valid
13:48:00JdGordonor almost never
13:48:02Torneunix allows anything except / and null
13:48:04Tornein a filename
13:48:08Torneand anything except null in a path
13:48:13Torneabsolutely any bytes you like :)
13:48:23JdGordonor do I just ignore the first / because rockbox doesnt do relative paths?
13:49:20JdGordonactually... maybe it doesnt matter
13:49:39pixelmaTorne: S_a_i_n_t was talking about the alarm plugin which isn't as old I think
13:50:09Torneoh right
13:50:22TorneS_a_i_n_t: the alarm plugin requires the player be left on and sitting in the alarm plugin
13:50:36pixelmaalthough I'm a bit confused who is talking about what
13:52:03pixelma"This plugin is an alarm clock, which resumes a paused song at a given time. "
13:52:45Tornethe asynchronous conversation i'm having with amiconn is about the real alarm
13:52:52Torneand whether it is maybe broken on archos
13:53:03Torneand/or other hypothetical players that hate the alarm being set while they're on
13:53:14pixelmaFM/V2 Recorders
13:54:33Tornei also mentioned yesterday that i was confused about what the alarm plugin is for; i now understand that it's for targets taht have an RTC but not an alarm
13:54:44Torneso you leave the player on and the plugin resums playback at the relevant time.
13:54:57Tornethough it seems like you could do the same thing for non-rtc targets just by specifying a countdown time instead of an absolute time
13:55:09Tornei guess nobody's bothered though
13:56:52pixelmathere once was a patch to start recording at a given time and with a countdown timer on non-RTC targets, it even got committed but was very wastefully coded and reverted (maybe it had some bugs as well). A while later the developer more or less vanished and nothing happened with it anymore
13:57:31Tornefor the actual alarm there's two questions/issues: 1) do some targets need to be powered off to avoid the alarm doing something bad, and if so does it do that or is it broken 2) what happens on other targets if the alarm goes off when the player is still on, it appears to clear the alarm and then do nothing about it which looks dumb
13:57:51Tornei can't help but think the builtin alarm should do the obvious thing on hardware that generates an interrupt or some kind of pollable status
13:58:04Torneand that we should only force powering off if the hardware makes it impossible to do anything else
13:58:19Tornethe code to poll the status appears to already be there :)
13:59:16Tornedoes anyone still have a v2/fm recorder and can actually put a current build on it?
13:59:25Tornebecause that'd answer that easily
14:03:00 Quit anewuser ()
14:03:18pixelmaI believe scorche has some but still being abroad
14:03:24 Quit Zarggg_ (Quit: Zarggg_)
14:03:24 Quit antil33t (Read error: Connection reset by peer)
14:03:32 Join kugel [0] (~kugel@rockbox/developer/kugel)
14:03:33 Join antil33t [0] (
14:04:12scorchei just returned today, however I only have v1 devices
14:04:27pixelmaoh, misremembered then
14:04:44pixelmamaybe LinusN... ;)
14:05:09scorchecheck older DevCon registries ;)
14:06:25pixelmaI even tried LinusN's FM-Recorder at DevCon 2007. Of course I don't know if he still has it and if it is in working condition :)
14:07:24kugelJdGordon: firmware/common/rbpaths.c
14:07:33JdGordonyeah, too slow.. found it :)
14:08:14LinusNi still have it, but I haven't used it since 2007 :-)
14:08:47pixelmaI wonder if the battery survived that
14:09:02LinusNi have a new battery for it somewhere...
14:09:28LinusNhehe, perhaps a year old or so
14:09:33 Join Zarggg [0] (
14:12:14 Join utrack [0] (
14:13:16 Quit utrack (Client Quit)
14:14:02CIA-7New commit by 03teru (r28267): fix sorting of track list when tracks with track number and tracks without track number are in the same list.
14:14:14 Join utrack [0] (
14:15:12 Quit utrack (Client Quit)
14:15:39CIA-7r28267 build result: All green
14:19:09 Join utrack [0] (
14:21:20pixelmahuh, if I load the rockbox_default.wps I get a blank screen except statusbar now?
14:22:20JdGordonthey was renamed
14:23:01pixelmaok, that works
14:23:45pixelmaDid I miss an announcement to delete the other one?
14:24:11JdGordona month ago
14:25:27pixelmawell yes, the old one won't be replaced anymore though and since I don't use it...
14:25:47 Quit jfc (Read error: Connection reset by peer)
14:26:12 Join jfc [0] (
14:26:16 Quit Loto (Quit: Loto)
14:26:22JdGordonmeh, something worth mentioning in the release notes maybe
14:27:10 Quit utrack (Remote host closed the connection)
14:27:33 Join Loto [0] (~ctrlproxy@xbmc/user/Loto)
14:28:24 Join utrack [0] (
14:29:48pixelmaLinusN: if you can spend some time testing a current build regarding behaviour if you set the alarm (if it'll immediately switch off or not), it could clear up some confusion :)
14:32:02 Quit Rob2223 (Quit: Rob2223)
14:32:20kugel <- test_codec results of my phone with and without asm enabled
14:33:34*pixelma wonders why there are rockbox_failsafe.rfms/rsbs/rwps on non-remote targets - is it also leftover?
14:33:50JdGordonno, just crap scripting
14:33:57kugelJdGordon: your code is broken! :P
14:35:18 Quit dfkt (Quit: -= SysReset 2.53=- Sic gorgiamus allos subjectatos nunc.)
14:36:16kugelI made my own test codec parser script which doesn't only output it in wiki format, I wonder if anyone is interested?
14:37:42JdGordonkugel: maybe its only the skin engine paths, but things shouldnt be calling get_user_file_path except for actual files
14:39:01kugelI don't understand
14:39:31JdGordonnever mind, that didnt entirely make sense
14:40:04JdGordonthings that store directory+filename shouldnt be calling get_user_file_path() to get the folder first, then append the filename
14:41:13*JdGordon would also like to change the sdl app to set the search path to <path> if rockbox is started like /<path>/rockbox
14:41:18JdGordonmake it more portable
14:41:35JdGordonfirst things first, get this working
14:42:54kugelit bugs me that mp3 isn't any faster
14:43:43pixelmakugel: meant to ask. There was a report that in the D2 sim the mapping of the arrow buttons to the top/right/bottom/left middle grid mode actions doesn't work anymore. It looked like something changed with your RaaA commits - has this been overlooked or did the reporter do something wrong?
14:44:28kugelI noticed that too but it worked when I began the android port (i.e. when I stopped working on the sdl app)
14:44:58kugelIIRC JdGordon made a commit to the buttons after that
14:45:22pixelmayou removed some files that looked related in r27310
14:46:16Zagorkugel: surely that qualifies as beast slaying?
14:46:30 Quit utrack (Remote host closed the connection)
14:46:55LinusNpixelma: i'll try to do that when i get the chance
14:46:59kugelZagor: I don't know the results of the beast for the other codecs, but for ape the beast is faster
14:47:22pixelmaLinusN: thanks
14:47:39kugelif I build for armv6 ape is a *lot* faster (2x as fast)
14:48:04Zagorhmm, why is there no beast in CodecPerformanceComparison
14:48:46 Join utrack [0] (
14:51:30 Quit krazykit (Quit: bye)
14:52:18 Join evilnick_B [0] (0c140464@rockbox/staff/evilnick)
14:54:07 Quit utrack (Read error: Connection reset by peer)
14:54:33CIA-7New commit by 03teru (r28268): fix button maps and assign some buttons for uisimulator.
14:56:09CIA-7r28268 build result: All green
14:57:08kugel"#ifndef CONFIG_PLATFORM" ? no wonder it didn't work
14:57:15kugelCONFIG_PLATFORM is always defined
14:59:47JdGordonget_user_file_path() really should be able to work without needing a buffer passed in
15:00:12kugelZagor: in an armv6 build my phone is a bit faster (for ape c5k)
15:00:33 Quit antil33t (Read error: Connection reset by peer)
15:00:44 Quit xsloader (Remote host closed the connection)
15:00:45 Join antil33t [0] (
15:01:33kugelJdGordon: that's to make it thread safe. non-main-threads are also doing file i/o
15:03:02JdGordonthis still isnt as transparent as it should be... font: /<Rb>/fonts//usr/local/share/rockbox/fonts/2.fnt
15:05:30***Saving seen data "./dancer.seen"
15:07:15 Join t0rc [0] (~t0rc@unaffiliated/t0rc/x-5233201)
15:08:19JdGordonoh woops.. cabbie.cfg is utterly broken
15:08:43 Join komputes [0] (~komputes@ubuntu/member/komputes) uses $rbdir which is coming from the Makefile.. problem is I want that to be /<Rb> and $rbdir to stay what it is
15:12:00 Part Zagor
15:12:22 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
15:12:42 Part LinusN
15:15:59 Part Zagor
15:16:06 Join LinusN [0] (~linus@rockbox/developer/LinusN)
15:16:43 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
15:18:24 Join Kitr88 [0] (
15:21:32 Quit Kitar|st (Ping timeout: 265 seconds)
15:22:44 Quit Kitr88 (Ping timeout: 240 seconds)
15:22:51 Join Kitar|st [0] (
15:27:12*TheSeven wonders if someone forgot to bump the plugin api recently
15:27:32 Quit petur (Ping timeout: 265 seconds)
15:27:37 Part LinusN
15:27:47TheSevenwhen i fixed that USB bug yesterday, i updated to a new rockbox binary, but didn't update the plugins
15:28:06TheSevennow i'm not getting any api incompatibility messages but instead Weird Behavior (tm)
15:28:30TheSevensome plugins work, some lock up with only a backdrop, and chopper is suffering from screen corruption
15:28:52TheSevenis that a known issue?
15:31:03*TheSeven wonders if anybody is awake
15:32:24terumaybe kugel's changes for display api?
15:33:00TheSevenchopper keeps drawing some random (and changing) mesh of green lines all over the screen
15:35:29 Quit tchan (Quit: WeeChat 0.3.3-dev)
15:36:57kugeldid I forget to update the plugin api?
15:37:08TheSeveni don't know
15:37:24TheSeveni just saw what the plugins are doing on my ipod :)
15:37:48kugelnone of my recent commits looks suspicious though
15:38:31terur28239 changes struct screen.
15:38:42kugelmaybe r28188 or r28102?
15:39:00kugelthat needs a plugin api bump?
15:39:15kugelhm right it probably does
15:39:47 Join Rob2222 [0] (
15:40:23terui think so. it is exposed to plugin api. and comment in screen_access.h tells to bump plugin api version.
15:42:00terur28188 doesn't affet since if HAVE_DIRCACHE isn't defined, build fails.
15:50:59 Part Zagor
15:51:35 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
15:53:01 Quit teru (Quit: Quit)
15:53:47 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
15:57:05 Join kevku [0] (
16:00:13hondza14:32 <kugel> <- test_codec results of my phone with and without asm enabled
16:00:27hondzakugel: do the new optimizations affect sansas as well?
16:00:38kugelthat's no new optimizations
16:00:42kugelit's enabling the existing ones
16:01:45 Quit tchan (Quit: WeeChat 0.3.3-dev)
16:04:51 Quit Zagor (Read error: Connection reset by peer)
16:06:38 Join Guest30082 [0] (~bjst@rockbox/developer/Zagor)
16:07:36 Nick Guest30082 is now known as Zagor (~bjst@rockbox/developer/Zagor)
16:08:13 Join jgarvey [0] (
16:15:01 Quit drizztbsd (Ping timeout: 260 seconds)
16:15:49 Quit JesusMcCloud (Ping timeout: 240 seconds)
16:15:49 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
16:24:02amiconnS_a_i_n_t: The alarm plugin is not supposed to shut down the device. We were talking about the built-in alarm
16:25:38 Quit Judas_PhD (Quit: This is a quitting message)
16:25:42amiconnJdGordon: < and > aren't valid on fat, that's why we chose it for multivolume as well ( /<MMC1>/folder/name )
16:28:50 Join drizztbsd [0] (~quassel@unaffiliated/drizztbsd)
16:31:03 Join leavittx_ [0] (~lev@
16:31:33amiconnkugel: armv6 ape filter code is expected to be much faster than the armv5 version, because it makes use of armv6 simd (packed 16 bit operands in standard 32 bit arm registers)
16:31:44 Join T44 [0] (
16:35:24kugelamiconn: we're not doing fat on raaa
16:36:42 Join Judas_PhD [0] (
16:37:11 Quit Judas_PhD (Client Quit)
16:39:31amiconnNo, but (1) rockbox usually obeys fat limitations because all real targets use it and (2) even if those chars may be valid on other file systems, they're rather unlikely to be used, because of their special meaning
16:41:10 Join Judas_PhD [0] (
16:42:52 Quit Judas_PhD (Client Quit)
16:44:59 Join panni_ [0] (
16:47:23 Join Judas_PhD [0] (
16:49:26 Quit Judas_PhD (Client Quit)
16:54:11 Part Zagor
16:55:31 Join Judas_PhD [0] (
17:05:31***Saving seen data "./dancer.seen"
17:07:33leavittx_thanks for fixing simulator!
17:07:36 Quit kugel (Remote host closed the connection)
17:09:51 Join sasquatch [0] (
17:15:09 Join dfkt [0] (dfkt@unaffiliated/dfkt)
17:17:47 Quit panni_ (Read error: Connection reset by peer)
17:29:32 Join JesusMcCloud [0] (~JesusMcCl@
17:31:16 Part pSiKO
17:31:32 Join G4Oblivion [0] (
17:34:58G4OblivionI think this was reported a while back, I thought it was fixed but when I change my theme to rockbox-failsafe the last themes backdrop and icons are still there.
17:38:13 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
17:38:55 Join DSStrife89 [0] (~Strife89@
17:40:56 Quit t0rc (Remote host closed the connection)
17:43:52 Quit DSStrife89 (Read error: Connection reset by peer)
17:45:03 Join DSStrife89 [0] (~Strife89@
17:49:15 Quit G4Oblivion (Quit: ChatZilla 0.9.86 [Firefox 3.6.12pre/20101013074404])
18:02:15 Join saratoga [0] (9803c22e@gateway/web/freenode/ip.
18:03:39saratogaso i'm reading through the cortex A8 instructions, and I'm baffled at why this one is useful: "doubling long multiply", which apparently does a long multiply and doubles the product
18:03:45saratogawhat am i missing
18:04:59Torne"DSP stuff", i'm guessing
18:05:04Tornesame as UQADDSUBX
18:05:05Torneor whatever
18:05:10 Quit DSStrife89 (Ping timeout: 276 seconds)
18:05:22saratogai see
18:05:33saratogatheres an option to just return the high word
18:05:53saratogaso you can do a s.31*s.31 fixed point multiply in a single op
18:06:58saratogasince doubling the high word is the same as doing ((int64)(int32x*int32y)>>31)
18:07:16saratogait would be nice if the manual could give some hint what these things are for!
18:08:07saratogabut anyway, neon looks kind of amazing for audio
18:09:30Torneneon is very well designed, yes, becuase they got to rip off the last quiteafew years of SIMD stuff on x86 :0
18:09:55saratogai like that the instruction codes are actually comprehensible
18:10:12Tornethe ARMv5E stuff is much less rational :)
18:10:34saratoga5e has basically nothing at all useful for audio as far as I can tell
18:10:42saratogajust some packed ops for video
18:10:45 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
18:10:51Torneno idea.
18:11:46saratogaone thing thats not clear to me about NEON is how it shares memory with the main processor
18:12:10saratogaam i correct to think that i can basically load anything into it and store anything back to cache, and also take things from the arm core cheaply
18:12:23saratogabut that i should probably not try to load things from neon into the arm core without first storing them?
18:12:28 Quit lestatar (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716])
18:13:47Tornenot sure what you mean..
18:13:54Tornethe NEON is part of the core
18:14:00Torneit's not a coprocessor in any meaningful sense :)
18:14:23saratogaon the A8 its technically a coprocessor
18:14:24Torneas far as i know it has exactly the same connection to the cache as the ALU does
18:14:35saratogaso you have to use MCR instructions to get stuff to it from the main register file
18:14:51Tornethat doesn't make it a coprocessor
18:14:59Tornethat's just how ARM choose to extend the instructoin set space nicely
18:15:15Torneit's tightly integrated with the CPU
18:15:41Torneshares the same cache interface and all that.
18:15:57saratogaits physically integrated, but logicaly its a coprocessor and theres apparently all kinds of nasty stalls involved with transfering stuff between the main core and neon
18:17:03 Join user890104 [0] (
18:17:23Tornemcr/mrc is nasty, but via the cache should be uninteresting..
18:17:36Tornethey're both driving the same cachecontroller
18:17:53Tornethat was what i meant
18:18:48Tornemrc and mcr both suck pretty badly
18:19:11saratogathe impression i get is that you mostly use NEON for jobs that do "gather, ADD/MUL/wahtever, scatter" type operations without any real flow control
18:19:34saratogawhile the main CPU is figuring out branches based on some hopefully independent set of variables from what neon is doing
18:19:41saratogais that more or less correct
18:19:53Tornei think so.. people use it as just a clever fpu as well though
18:20:12saratogathe load/store instructions look amazing
18:20:25Tornethat too; people use it for memcpy() :)
18:20:30saratogaperfect for filterbanks like mp3's
18:26:34saratogai wonder how much difference optimization really makes on android
18:26:44saratogacan we do battery benches on it yet?
18:28:03 Join wodz [0] (
18:30:37 Quit user890104 ()
18:30:42 Join user890104 [0] (
18:34:24 Quit saratoga (Quit: Page closed)
18:35:03 Quit swilde (Quit: ERC Version 5.3 (IRC client for Emacs))
18:36:09 Join user890104_ [0] (~Venci@2001:0:4137:9e74:8a:f6b:2b16:78b5)
18:36:17 Quit user890104 (Read error: No route to host)
18:36:38 Nick user890104_ is now known as user890104 (~Venci@2001:0:4137:9e74:8a:f6b:2b16:78b5)
18:38:20 Join user890104_ [0] (
18:41:17 Join Xerion [0] (
18:41:20 Quit user890104 (Ping timeout: 272 seconds)
18:42:11 Join user890104 [0] (
18:42:14 Quit user890104_ (Read error: No route to host)
18:42:32 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
18:47:52 Join Jerom [0] (~jerome@
18:54:16 Join _s1gma [0] (~d.d.derp@
18:58:34 Join DSStrife89 [0] (~Strife89@
19:00:54 Quit Jerom (Quit: Leaving.)
19:01:49 Join Jerom [0] (~jerome@
19:05:34***Saving seen data "./dancer.seen"
19:06:16 Quit Jerom (Client Quit)
19:07:29 Join n1s [0] (
19:07:29 Quit n1s (Changing host)
19:07:29 Join n1s [0] (~n1s@rockbox/developer/n1s)
19:08:19 Join krabador [0] (
19:08:37 Quit DSStrife89 (Quit: Back later.)
19:08:55 Join G4Oblivion [0] (
19:10:50bertrikHow good is fm reception on the gigabeat s?
19:10:51 Quit DerPapst (Quit: Leaving.)
19:11:43n1sbertrik: i think it's pretty good, not that i use it beyond testing
19:11:44bertrikI've seen a few complaints about the tuner chip used in the beast (si4700) not switching to stereo with relatively weak signals
19:12:04bertrikin comparison with the OF
19:12:11n1sah, never used the of
19:12:31n1sbut i thought that was an auto thing in the chip, no?
19:12:39 Join Topy [0] (
19:13:41bertrikn1s, it is possible to control the signal strength range where it transitions from mono to stereo
19:14:05bertrikso we could make it stereo for weaker signals, but with possibly more noise
19:14:43n1swell, fm nad the compression used by radio stations means everything sounds like crap anyway :)
19:15:05 Quit G4Oblivion (Quit: ChatZilla 0.9.86 [Firefox 3.6.12pre/20101013074404])
19:16:27 Quit T44 (Ping timeout: 265 seconds)
19:16:35n1shmm, in fact the gigabeast uses a different way of detecting stereo than the others with similar fm chips
19:17:47bertrikoh? Do you mean it checks a GPIO instead of checking a bit in a register?
19:18:23 Quit CGL (Remote host closed the connection)
19:19:38bertrikI think it's a si4701 in the beast instead of an si4700 as I said earlier
19:20:51n1syeah, it checks the gpio
19:23:16n1sso you think we should change BLNDADJ to indicate stereo for a lower signal strength?
19:24:05n1swe could make it configurable of course
19:24:14bertrikYes, possibly. I'm waiting for some feedback from users (both at FS and ABI)
19:24:45n1sah, yes si4701 is the one with rds, which iirc the beast chip has
19:24:55n1ssomeone should implement support for that :)
19:25:31bertrikProbably not that hard to do, just needs to be done
19:26:40n1syeah, the apps code should be there already since that ipod fm remote has rds support in rockbox
19:27:07bertrikMaking BLNDADJ user configureable sounds a bit overkill to me to be honest
19:28:08bertrikthe ipod fm remote rds support feels a bit hacked to me, it only supports the station name (8 chars) and the scrolling text thing (64 chars IIRC)
19:28:18bertrikbut it's a nice start to get something going quickly
19:28:30n1swe could pop it down to -6dB and if anyone wants mono, there's the option to force mono anyway
19:28:46*linuxstb was about to ask if there was an option to force mono
19:28:55n1sthere is
19:34:00n1shmm, that reminds me, i wanted to fix that frequency drifting thing with fm on the c200v1...
19:35:19bertrikI made a patch to set it to 6 dB lower but the bug submitter said it was still not OK yet compared to the OF :
19:36:17bertrikI realize now that perhaps the regional setting can cause problems too, e.g. it may result in tuning to a channel off by 0.1 MHz
19:36:53bertrikthat c200v1 tuner looks like hell to tune, compared with the newer tuner types
19:38:09n1snever looked at it, i just remember my mother used her c200 this summer and told me about the problem and it was very reproducable, in warm weather, it seemed very temerature dependant :)
19:38:52 Join Horscht [0] (
19:38:52 Quit Horscht (Changing host)
19:38:52 Join Horscht [0] (~Horscht@xbmc/user/horscht)
19:39:03n1sbut i figured that if it can read back the actual freq, it can retune after drifting too much or just retune after a set amount of time, assuming that doesn't cause other glitches
19:40:05bertrikWIth the newer tuner chips, tuning basically consists of setting a channel number and you're tuned within 100 ms or so. The c200v1 tuner requires a multi-step tune procedure where you need to calibrate the frequency using accurate delays.
19:41:29n1snothing wrong with tuning on it afaict, as chaning channel and back again fixed it but it seems to drift from the tuned freq as time passes so you get more and more static
19:41:49n1sand it drifts faster when it's warm :)
19:42:08n1sit is a reported bug in our tracker
19:42:37bertrikif you can measure it, you could use it as a thermometer :)
19:42:46n1sgood idea!
19:43:20n1sFS #9420, seems to be the same on e200v1
19:48:00bertrikWe can make a very crude thermometer by measuring the voltage difference between two diodes that run at different currents in the codec chip. In theory, the voltage difference is directly proportional to absolute temperature. The accuracy of the A/D is a bit too low to make a good temperature measurement though (or we'd have to average a lot of samples).
19:48:14bertrik^- talking about the c200v1 and the e200v1
19:48:56 Quit markun (Read error: Connection reset by peer)
19:52:21bertrikn1s, one of the things I'd really like to add eventually is a signal strength indicator for fm radio, most fm chips actually seem capable of reporting this (with varying accuracy)
19:54:06n1sbertrik: not a bad idea
19:59:03 Join Staphylo [0] (
20:00:30 Quit krabador (Ping timeout: 272 seconds)
20:05:08 Join pamaury [0] (
20:05:09 Quit pamaury (Changing host)
20:05:09 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
20:06:09pamaurygevaerts: do you know how the bootloader works with respect to usb ? It seems that with my patch on FS, the bootloader doesn't compile anymore (seems like usb.c is not compiled in), do you have an idea ?
20:06:18 Join DerPapst [0] (
20:06:50pamaury(the question is partly specific to amsv2 and partly generic, there are errors of both kinds it seems)
20:08:02 Join clone4crw [0] (
20:09:13 Quit dfkt (Quit: -= SysReset 2.53=- Sic gorgiamus allos subjectatos nunc.)
20:12:37 Quit _s1gma (Ping timeout: 265 seconds)
20:14:24gevaertspamaury: not really, no. All bootloaders I ever dealt with do no usb :)
20:14:30linuxstbpamaury: I think in general the bootloaders don't have the full USB driver. On PP ipods (and I'm guessing other PP devices) we don't even enable interrupts.
20:16:08pamauryI don't know, the as3525 does call to usb_start_monitoring and usb_core is built and require usb_signal_transfer_completion...
20:24:37 Quit DerPapst (Read error: Connection reset by peer)
20:26:24 Quit fyrestorm (Read error: Connection reset by peer)
20:28:42 Join _s1gma [0] (~d.d.derp@
20:29:15 Join DerPapst [0] (
20:36:45 Quit Topy (Ping timeout: 252 seconds)
20:38:16 Join krabador [0] (
20:38:56 Join krazykit [0] (
20:41:24 Join T44 [0] (
20:42:06 Join markun [0] (
20:42:06 Quit markun (Changing host)
20:42:06 Join markun [0] (~markun@rockbox/developer/markun)
20:58:21clone4crwHey could someone review does call to usb_start_monitoring and usb_core is built and require usb_signal_transfer_completion...
20:58:24clone4crw13:24 -!- DerPapst [] has quit [Read error: Connection reset by peer]
20:58:27clone4crw13:26 -!- fyrestorm [] has quit [Read error: Connection reset by peer]
20:58:30clone4crw13:28 -!- _s1gma [~d.d.derp@] has joined #rockbox
20:58:33clone4crw13:29 -!- DerPapst [] has joined #rockbox
20:58:35clone4crw13:36 -!- Topy [] has quit [Ping timeout: 252 seconds]
20:58:39clone4crw13:38 -!- krabador [] has joined #rockbox
20:58:42clone4crw13:38 -!- krazykit [] has joined #rockbox
20:58:45clone4crw13:41 -!- T44 [] has joined #rockbox
20:58:47clone4crw13:41 -!- markun [] has joined #rockbox
20:58:50clone4crw13:41 -!- markun [] has quit [Changing host]
20:58:53clone4crw13:41 -!- markun [~markun@rockbox/developer/markun] has joined #rockbox
20:59:13krabadorclone4crw, are you ok?
20:59:15*clone4crw is very sorry for that
20:59:26clone4crw* was having trouble with pasting in rxvt
21:02:16clone4crwWhat I meant was could someone review FS #11667 for me? (I'm too scared to try pasting again)
21:05:20gevaertsclone4crw: I'm trying to understand why it's needed
21:05:36***Saving seen data "./dancer.seen"
21:07:30clone4crwBecause 150 ohms was showing up as brown - green - black. The last one should be brown, it happens only between 100 and 199.
21:07:48gevaertsYes, I believe it's wrong, but why?
21:10:02 Quit krabador (Ping timeout: 265 seconds)
21:10:54clone4crwBut why... is it a problem in the first place?
21:10:57 Quit linuxguy3 (Read error: Connection reset by peer)
21:11:32gevaertsI mean, isn't there a bug elsewhere that this patch would just hide?
21:12:34clone4crwYes, I know this is a kludge. I haven't had time to review it and really fix it.
21:13:05clone4crwIt just bugged me when I went to actually use the plugin
21:13:13gevaertsYes, I understand that :)
21:13:36gevaertsActually, is the patch correct? I mean, if the value is excatly 100, what should happen?
21:14:08clone4crwI believe (i don't have it open now), that this applies if its > 100 and <= 199
21:14:14gevaertsso not for 100?
21:14:20clone4crwNope. that's fine.
21:14:48 Join linuxguy3 [0] (
21:15:09clone4crwStrange, I know. Next chance I'll get I'll look for a real solution.
21:15:21 Join krabador [0] (
21:17:17*gevaerts tries some numbers
21:19:50gevaertsclone4crw: it looks like with the svn version 109 is still correct, 110 is wrong
21:20:52amiconnThe whole range handling in that plugin looks strange and error prone to me
21:23:08clone4crwwhat's the difference between the two versions?
21:23:50gevaertsclone4crw: I mean, it looks like your patch is wrong. 101 to 109 *don't* need fixing
21:26:39*gevaerts looks at the code in detail
21:27:11clone4crwoh. line numbers. gotcha.
21:27:30amiconnNo, values (iiuc)
21:27:43*clone4crw is a bit confused
21:27:49gevaertsclone4crw: try entering 105 in the resistance-to-colour part
21:28:41 Join MethoS- [0] (~clemens@
21:30:12amiconnHmm, also the plugin doesn't handle multipliers with negative exponent
21:30:27clone4crwBrown black red sure isn't 105 ohms.
21:31:47amiconnAnd it doesn't know tolerances <1%
21:33:22clone4crw110 seems to be correct, though.
21:39:16gevaertsclone4crw: at first sight, seems to give me a correct resistor to colour conversion
21:40:51clone4crwI'll give it a try.
21:41:22gevaertsI'm now doing the same in the LED code
21:43:56gevaertsclone4crw: I've put my patch on FS #11667
21:44:12gevaertsI *think* it's correct
21:45:42amiconngevaerts: You can make that even more efficient if you just calculate tens=kbd_input_int%100; in the loop
21:46:18amiconnAfter the loop: units=tens%10; tens /= 10;
21:47:08gevaertsOh, indeed
21:47:08 Join s1gma_ [0] (~d.d.derp@
21:48:17gevaerts would then be the result
21:48:19amiconnHmm, if you switch kbd_input /= 10; and the if(...) break; (which in turn needs to check for < 100 then), you don't even need that
21:48:52amiconnYou would just calculate units and tens after the loop from kbd_input_int
21:49:21gevaertsamiconn: this isn't the inner loop of a codec :)
21:49:59amiconnOf course not, but shorter code is easier to understand...
21:50:20 Quit _s1gma (Ping timeout: 240 seconds)
21:53:10amiconnJust the loop:
21:59:41 Join stripwax [0] (
22:13:46CIA-7New commit by 03gevaerts (r28269): Simplify resistance to colour calculations, removing some bugs in the process. ...
22:15:13CIA-7r28269 build result: All green
22:15:27gevaertsThe temporary variable wasn't technically needed for the second case, but the LED resistor calculator still needs the value later on, and I preferred to keep the code similar
22:18:07amiconnIf the original variable isn't used afterwards, gcc will even generate the same code as if you used the original directly
22:18:42clone4crwAnd I was about to say that the patch seems to be working
22:19:10gevaertsclone4crw: I tested some values too :)
22:24:53 Quit stripwax (Quit:
22:25:03clone4crwI'm recompiling with r28269 now.
22:30:34amiconngevaerts: Looks like you forgot something... int get_power_ten(int in_val) can be removed altogether
22:31:16*amiconn wonders why all these helper functions aren't static
22:31:29amiconnIf they were, gcc would warn if they're unused
22:31:37gevaertsgood point
22:36:01CIA-7New commit by 03wodz (r28270): HD200 - GPIO57 is headphone output enable. Proper handling of this line fixes hissing noise after startup.
22:37:53CIA-7r28270 build result: All green
22:38:48bertrikamiconn, lots of functions in plugins could be made static, I won't mind working on cleaning that up
22:39:01bertrikOTOH, it doesn't seem to matter *that* much
22:42:00gevaertsThere are more bugs. Black-brown-red isn't 100 kilo ohm
22:43:22CIA-7New commit by 03gevaerts (r28271): Resistor calculator: Add lots of static, and slightly reorganise the code
22:43:56 Quit kevku (Quit: KVIrc 4.0.2 Insomnia
22:45:11CIA-7r28271 build result: All green
22:52:02 Quit krazykit (Ping timeout: 255 seconds)
22:52:17 Join stripwax [0] (
22:53:04 Quit Barahir (Ping timeout: 252 seconds)
22:53:36 Quit thegeek (Quit: ( :: NoNameScript 4.2 :: ))
22:53:49 Join thegeek [0] (
22:54:51 Quit stripwax (Client Quit)
23:03:41gevaertsclone4crw: it seems that technically, violet, grey and white are valid multipliers too. Shouldn't we add those as well for completeness?
23:04:36*gevaerts realises that 99 gigaohm resistors aren't very common...
23:05:39***Saving seen data "./dancer.seen"
23:06:03CIA-7New commit by 03gevaerts (r28272): Fix some incorrect colour-to-resistance results. The price is that sometimes the display is subobtimal (like "2000 Ohms" instead of "2 KOhms") ...
23:06:39clone4crwhaha. If I knew what those multipliers were, sure. Why not?
23:07:31gevaerts lists them
23:07:38CIA-7r28272 build result: All green
23:07:42gevaertsThey just follow the normal order
23:08:02*gevaerts will add them
23:11:21 Quit bmbl (Quit: Bye!)
23:11:29 Quit evilnick_B (Quit: Page closed)
23:12:27CIA-7New commit by 03gevaerts (r28273): Add violet, grey and white multiplier for completeness
23:13:00gevaertsI think it's bugfree now
23:13:33gevaertsWell, except for the "1000 Ohms" instead of "1 KOhms" thing, but I prefer that over wrong answers anyway :)
23:13:50CIA-7r28273 build result: All green
23:16:08clone4crwI agree, in the resistance to colors mode, if the unit is 'ohms', then I kind of want the answer to be 'ohms'
23:16:41gevaertsThe best solution for that really is to keep everything in tens, units and power
23:17:35gevaertsThen that sort of thing is trivial, but other calculations may not be any more. I suspect the LED calculator would get harder to understand
23:22:25amiconngevaerts: White multiplier (also grey for some values) will overflow 'int'
23:23:56gevaertsamiconn: in which cases? Colour-to-resistance should handle that
23:24:04gevaertsThe other direction may not
23:24:23amiconnThe latter
23:25:03 Join zdebel [0] (
23:25:11gevaertsWe'll just give a different wrong answer now though. I think it used to return black-black-black for big enough values
23:25:12zdebelhi, anyone with a gigabeat s in the house xD?
23:25:21zdebelI've been using rockbox for like over 1 year now
23:25:22Tornei have one in the house, yes
23:25:26zdebeland today something WEEEEEEEIRD happened
23:25:28Tornenot sure where exactly
23:25:36zdebelit fails to read the partition x.x
23:25:38zdebelout of the blue
23:25:49Torneyou mean the bootloader?
23:25:50Torneor what?
23:25:52zdebeland atm I'm at the firmware restore part, the original one
23:26:05zdebeland fucking updater won't discover the device
23:26:13Tornedo you mean our bootloader errored, or do you mean the original firmware sent you to the restore process?
23:26:28zdebelboth mate
23:26:37zdebelat first it was working, I jsut reformated the data partition
23:26:41zdebelput a brand new .rockbox in
23:26:47Tornethe bootloader complained and then when you restarted the OF prompted you to restore?
23:27:16Tornewell, the beast is.. fun like that :)
23:27:20zdebelit wouldn't be a biggie if the updater app found the device... x.x
23:27:27 Quit domonoky (Read error: Connection reset by peer)
23:27:39Torneare you using the toshiba update utility?
23:27:42zdebelwindows (sadly 7) sees it properly
23:27:46Torneyou could use our firmware downloader instead
23:27:50Tornethat might work better
23:27:57zdebelhaven't heard about it
23:28:15*zdebel checks the site then!
23:28:18Tornealso there should be a version 1.3 of the toshiba one too
23:28:35Tornewhich may work better perhaps
23:28:37zdebelok I'll first try to find that then
23:28:47 Quit wodz (Quit: Leaving)
23:28:56Torne <- this has a link to sendfirm.exe which is our MTP firmware download tool
23:29:06Torneyou'll need to get the right nk.bin to download, though
23:29:29Torneyou might be able to do it with rockbox utility as well, just by installing the rockbox bootloader again?
23:29:36Tornesince that uses the exact same process
23:29:55zdebelI'll see what I can do then
23:30:24zdebelI've been using rockbox for so long now, loving every part of it every day
23:30:28zdebeland now this happens, out of the blue
23:30:32zdebeleh my luck recently
23:30:42CIA-7New commit by 03gevaerts (r28274): Resistor calculator: Use correct units if resistor to colour uses kilo- or megaohms
23:30:43Torneunfortunately it may just be the beast firmware doing the thing that it does
23:31:05Tornewe still haven't figured out why it sometimes decides the disk is corrupt and reformats it for you :)
23:31:15zdebelyeah, I've read that on the page
23:31:16 Quit S_a_i_n_t (Disconnected by services)
23:31:19 Join S_a_i_n_t [0] (S_a_i_n_t@
23:31:22zdebelwas like, that never happened to me o.o
23:31:24zdebeland then...
23:31:32Torneit happens very frequently for some people :)
23:31:43Tornealternatively, maybe something else is wrong; could be that your disk is dying or something, I guess
23:31:59Tornebut if you can restore it and it appears to work afterward then i'd just blame the beast boot code
23:32:00zdebelwas working perfectly fine, didn't throw it at anything etc
23:32:08Tornedisks die eventually anyway :)
23:32:17zdebelso um, the rockbox utility supports the >s< version now?
23:32:23CIA-7r28274 build result: All green
23:32:55TorneOh, er, no
23:33:10zdebelok so first I'll try to find the official 1.3 updater
23:33:15zdebelif not, I'll sendfirm 1.2
23:33:58Tornebeastpatcher might be able to do the same thing as sendfirm, also
23:34:05*zdebel crosses fingers
23:34:23zdebelaw shit the updater still doesn't find the device
23:34:34n1sbeastpatcher is probably the more-likely-to-work one, no?
23:34:42Tornen1s: they're built from the same code, iirc
23:34:50Tornethough the sendfirm.exe linked to there probably wasn't built recently
23:35:02zdebelI've used it in the first place to put rockbox on this player
23:35:05n1soh, also has anyone tried either of those tools on windows 7?
23:35:15*zdebel extracts nk.bin
23:35:17Tornei'm sure someone has :)
23:35:29Tornei haven't actualyl done anything with the beast for a long time so not me
23:35:56n1sme neither, i don't reinstall the bootloader often at all
23:36:12zdebelok got the nk.bin, what do you suggest to try first?
23:37:25Tornetry just folling the normal rockbox manual install instructions
23:37:30zdebelI think it won't work since it uses the mtp protocol, which sux balls on windows 7
23:38:36Torneit doesn't use windows' implementation of mtp
23:38:43Torneit uses a library to do it
23:38:57Torneit may not work, i have no idea if it works on win7..
23:39:01Tornebut it might..
23:39:42zdebelbooted xp in vm
23:40:22Torneif you want to just reinstall the toshiba firmware for now you can do beastpatcher -s nk.bin on the toshiba nk.bin
23:40:32Tornewhcih does the same thing as sendfirm, just sends it as-is without patching it
23:42:27Torneor you can just follow the instructions and install a dual boot bootloader directly
23:42:51gevaertsamiconn: I think fixes that
23:43:13zdebelTorne, atm I wish >something< will manage to get installed
23:44:51gevaertszdebel: not that it helps, but this sort of thing is actually why we haven't declared rockbox on gigabeat S stable yet
23:45:05zdebelyeah I understand
23:45:26zdebelbut it's the 1st time smething like this happened to it SINCE I've been using it
23:45:33zdebelthus my :@ mood
23:45:46zdebelwhoa speech support :D
23:48:53zdebelbash: ./beastpatcher: /lib/ bad ELF interpreter: Nie ma takiego pliku ani katalogu
23:49:04*zdebel links /lib64's .so.2 to /lib for a sec
23:49:12zdebelwin xp of course failed to detect the device
23:49:20zdebelmight be vbox though
23:49:42Tornedid you already try it in windows 7? :)
23:49:53zdebelsendfirm yes, beastpatcher no
23:50:02Tornethat sendfirm binary is probably ancient
23:50:15Tornedid it prompt you to run it as administrator?
23:50:40Tornethe current beastpatcher has a manifest for that
23:50:49zdebelI ran it in cmd.exe
23:50:57zdebeland it didn't prompt me though
23:50:59 Quit bertrik (Quit: :tiuQ)
23:50:59zdebelmight be it!
23:50:59Torneright, but it still needs to run as administrator probably
23:51:05Torneand if it didn't prompt you then it won't've done :)
23:51:20 Quit jgarvey (Quit: Leaving)
23:51:41zdebellemme find a new sendfirm
23:51:42 Join krazykit [0] (
23:51:48 Join soap_ [0] (~soap@rockbox/staff/soap)
23:52:02Tornethere probably isn't one
23:52:07Torneits functionality is built into beastpatcher
23:52:13zdebeloh ok
23:52:14Torneso there's no need to build it any more
23:52:48Tornethey do the same thing, except for any changes in the time between the binaries being updated
23:52:48CIA-7New commit by 03gevaerts (r28275): Resistor calculator: Support Gigaohms in resistor to colour conversion, and guard against values that can not be represented
23:53:00Torneso if the current beastpatcher doesn't work then no version of sendfirm will either
23:53:09Tornethey are built from the same code, beastpatcher just also knows how to patch in the rockbox bootloader
23:53:13zdebellame question, where can I find the current one?
23:53:20Tornefrom the install instructions :)
23:53:37zdebelah ok, I got a feeling you meant they were outdated
23:53:55Tornesendfirm is probably outdated, yes
23:54:06Tornebtu the install instructions don't use it
23:54:11Tornethat was just from when the port was being developed
23:54:28CIA-7r28275 build result: 0 errors, 23 warnings (gevaerts committed)
23:54:37zdebelno devices found :(
23:55:25 Quit soap (Disconnected by services)
23:55:27 Nick soap_ is now known as soap (~soap@rockbox/staff/soap)
23:55:53 Join soap_ [0] (~soap@rockbox/staff/soap)
23:56:18zdebel>.> the linux version find it
23:56:24 Quit s1gma_ (Quit: Leaving)
23:56:38CIA-7New commit by 03gevaerts (r28276): Resistor calculator: Initialise variable to make the compiler shut up about it (it's always initialised, but the compiler won't know that)
23:56:40*zdebel transfers the nk.bin
23:57:00zdebelwhoa a resistor calculator, I'd love that xD!
23:58:15CIA-7r28276 build result: All green

Previous day | Next day