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 2007-10-24

00:00:44 Quit J3TC- (Read error: 110 (Connection timed out))
00:01:48 Join Echelon [0] (
00:02:47 Join Sedgewick [0] (
00:03:58 Join ddalton [0] (
00:04:08ddaltonwhat code is executed when the player is turned on?
00:04:37linuxstbThat depends on the player, but normally the code in flash ROM.
00:04:52linuxstbWhich may be part of the original firmware, or may be Rockbox...
00:05:18ddaltonlinuxstd: do you know what source file?
00:05:29ddaltonI want to call a function once rockbox has loaded
00:05:45preglowi can't make this bag of shit work
00:06:23linuxstbddalton: Are you asking which code in Rockbox is run first?
00:07:15ddaltonso what code resumes the playback for example on startup
00:08:29linuxstbThe main entry point is main() in apps/main.c
00:08:43preglowi've disabled all memory accesses and all loops, this has to be a really, really cool bug...
00:08:53ddaltonok thanks
00:09:43ddaltonlinuxstb: do you know how I can check if the alarm was activated? so the player booted because of the alarm not the user?
00:09:56linuxstbstripwax: Did you ever finish your runtime test in diagmode?
00:10:24jhMikeSpreglow: don't you just love those one where after you've removed all the code it still crashes? :P
00:10:46*linuxstb normallly then realises he's editing a different source tree to the one he's compiling...
00:11:21ddaltonso it is called "int main(void)"?
00:11:43jhMikeSyes, everything crashes in main() :)
00:11:58ddaltonwhat code is executed when the alarm wakes up the player?
00:12:41preglowthe same startup sequence runs
00:12:54preglowsome place in it you'll find a call to alarm code to check if an alarm was what triggered the powerup
00:13:00ddaltonso you can't tell if it is the alarm or not?
00:13:01jhMikeSthe alarm just turns on the power
00:13:06 Join webguest21 [0] (i=50b2489e@gateway/web/cgi-irc/
00:13:38webguest21hello, how can i add date to my main screen, it doenms't show up
00:13:45ddaltonjhMikeS: So I can't see if the alarm has woken it up?
00:13:46preglowwhat the HELL is going on
00:14:24linuxstbwebguest21: Do you mean the While Playing Screen (WPS) ?
00:14:25jhMikeSpreglow: sourcery
00:15:03jhMikeSddalton: I think preglow said you'll find that in the alarm code
00:15:13preglowi forgot to test another optimization, and kept it enabled :////////////7
00:15:16ddaltonjhMikeS: ok thanks
00:15:16jhMikeSddalton: iirc he wrote it
00:15:19jeplerhm is there something wrong with the latest svn? I updated (from r15097, nov 13 or so) and ogg playback is skipping quite a bit. the "drive" activity indicator is stuck on
00:15:33linuxstbwebguest21: What have you added to your .wps file, and what device are you running Rockbox on?
00:15:56webguest21i ruunnig e200.... what u mean added to wps?
00:16:03jeplerlogf shows that it's in the middle of updating the database, maybe that's the cause?
00:16:26preglowjhMikeS: i didn't write it, but i did port it to ipods
00:16:37 Join HellDragon [0] (n=Nocebo@unaffiliated/helldragon)
00:16:40linuxstbwebguest21: You said the date doesn't show, what did you do try try and make it appear?
00:17:29webguest21i just want to add the date to menu screen , is it possible?
00:17:59webguest21or it should be added on theme?
00:18:22linuxstbYou can add it to the WPS (which is part of a theme), but not to the menu screens.
00:18:45preglowit seems i was right about qmf_synth() being important
00:19:13webguest21ohhhh ok thanks
00:19:17preglowit doesn't sound quite right in asm, though...
00:20:05 Quit webguest21 ("CGI:IRC (EOF)")
00:20:34preglowjhMikeS: i think i'm about to make my first use of the emac internal shifting feature too
00:20:34 Quit japc (Read error: 104 (Connection reset by peer))
00:20:35 Join webguest14 [0] (i=50b2489e@gateway/web/cgi-irc/
00:20:40jepleralso I see that there's a change to bootloader/pp-main.c related to Rebuilding Database in OF. Do I re-run sansapatcher to make sure I have the new bootloader and the update to the "nvparams" that it mentions?
00:20:55ddaltonjhMikeS: hmmm I can't seem to find wake up code in alarm_menu.c. Had a quick read through the whole file. Who wrote it?
00:21:07webguest14is it possible to use one theme with other background image?
00:21:09preglowddalton: exactly what code are you looking for?
00:21:44ddaltonpreglow: the code that is ran when it wakes up. I want to adjust the volume if the alarm wakes up the player
00:22:32preglowgrep for "check_alarm_started"
00:22:48Zagorjepler: yes, you run sansapatcher again to update the bootloader
00:22:56amiconnlinuxstb: Do you know what arm version the pnx0101 is?
00:22:59preglowup to that point in the code (and for most code after it), the exact same code is run, no matter if the alarm set off or not
00:23:24amiconnProbably not very important, as it might be impossible to get ape realtime on that one...
00:23:30stripwaxwebguest14 : - yeah, just load a different background image ....
00:23:48linuxstbamiconn:: No idea, but tools/configure selects arm7tdmicc
00:24:03linuxstbI don't know the speed either...
00:24:07amiconnSO the only non-arm7 atm is the gigabeat
00:24:24linuxstbOuch... 60MHz and 1MB - not an easy job.
00:24:35ddaltonok I better go
00:24:37amiconnI wonder how to handle different optimisations for the various arm versions...
00:24:41webguest14stripwax: how i do it??
00:24:53webguest14i have e200
00:24:56preglowddalton: did you find what you needed?
00:25:05webguest14i can't do it
00:25:08webguest14i tried
00:25:26amiconnLooks like the vector add has no effect on arm9 (neither slower nor faster), but quite some speedup on arm7
00:25:28ddaltonthanks for the help but I better look later. Really got to go
00:25:32jeplerZagor: thanks, trying that now
00:25:42webguest14i can change background only when i changetheme
00:25:46amiconnBut for scalarproduct(), the arm7 optimisation would *slow down* arm9 (and higher)
00:25:51webguest14can someone help please?
00:25:57 Quit ddalton ("leaving")
00:26:30stripwaxwebguest14 - navigate to the background you want (using the File list), hold down the Select button / context menu and pick the "Set As Backdrop" option. Should be easy enough ..
00:26:45stripwaxassuming it's a .bmp that is. otherwise...
00:27:06linuxstbamiconn: I still don't understand ARM's versioning system - I know there are different architectures (v4, v5 and v6) with different instruction sets, but then we also have different performance within the same arch...
00:27:07stripwaxwebguest14 - you looked in the Manual at all?
00:27:39rasherstripwax: which reminds me - the fact that backdrops must be in .rockbox/backdrop (or whatever) is a limitation from the days of config-sector, isn't it?
00:28:00 Quit A_M ()
00:28:41stripwaxrasher - oof, wow, dunno. is that still true?
00:28:45webguest14ohhh i found it thanks, it in files folder
00:28:52amiconnlinuxstb: There's even more to it. T and non-T variants, various optional coprocessors etc etc
00:29:05linuxstbamiconn: Yes, I was simplifying ;)
00:29:07rasherstripwax: The manual still says so, and some guy came in here complaining about it, so I'd guess it is
00:29:33jeplerOK, where do I get the .mi4 files necessary to build sansapatcher?
00:30:05linuxstbamiconn: Maybe it's worth looking at how gcc categorises the processors. Or maybe for now, just use #ifdefs based on the processor - e.g. ARM7TDMI etc
00:30:16pixelmarasher: I think it's still the case and it's also a nice way to just try and return back to normal
00:30:23amiconnDo we have those?
00:30:35rasherpixelma: return back to normal?
00:30:42rasherAh.. like that
00:30:47Zagorjepler: those are the bootloaders for c200 and e200. so build those first.
00:30:57linuxstbamiconn: No, but they should be easy to add to config.h based on CONFIG_CPU
00:30:57 Quit Arathis ("Bye, bye")
00:31:19rasherWell, it just seems like a completely arbitrary limitation (now), which I thought we disliked
00:31:19Zagorjepler: or grab them from
00:31:57 Join Mouser_X [0] (n=someone@
00:32:37pixelmarasher: if you just want to try some backgrounds and then you realise that it has not enough contrast to your text colour
00:32:58rasherpixelma: yeah, but then you can just go into the menu and clear backdrop, surely?
00:33:33pixelmanot if you can't read the text anymore
00:33:51amiconnlinuxstb: Hmm. Telechips is arm v5te, correct?
00:33:51pixelmabesides there are other things that are stored in their own folders inside .rockbox like the fm and eqpresets
00:34:04amiconn(i.e. arm946e-s)
00:34:18rasherpixelma: those are less likely to be changed by users browsing to the files though, I'd imagine
00:34:35linuxstbamiconn: Yes, sounds right.
00:35:09jeplerZagor: that's not the same as firmware.mi4 right?
00:35:30Zagorjepler: you need both, since sansapatcher is used for both e200 and c200
00:36:09 Quit bluebrother ("leaving")
00:36:52*amiconn wonders why arm9 has both members with armv4t and armv5te architecture...
00:37:31 Quit webguest14 ("CGI:IRC (EOF)")
00:37:34jeplerer, that's not the same as rockbox.mi4, right?
00:37:38pixelmarasher: I don't have a strong opinion on that, just saying that there are upsides as well. But I noticed that there is one big difference to the fm and eq presets - you can browse the latter through the menu, whereas changing the backdrop if it's inside that folder needs the file view set to all
00:37:44jeplerit's some part of the OF of those devices?
00:38:39rasherpixelma: Neither do I - it just seems weird that you can change backdrop, but for some unknown (to the inexperienced user) reason, the setting doesn't stick
00:38:49 Quit jhMikeS (Nick collision from services.)
00:38:50linuxstbZagor: Did you read my discussion the other day with bertrik about my Telechips usb problem with Linux?
00:38:55 Join jhMikeS [0] (n=jethead7@rockbox/developer/jhMikeS)
00:38:57jeplerwell .. I just used a new sansapatcher binary from and the rebuild disable seems to work for me now
00:39:11Zagorlinuxstb: no, got a link/time?
00:39:22Zagorjepler: excellent
00:39:33linuxstbZagor: 21 October, starting at 13.13.01
00:40:49Zagorlinuxstb: have you tried usbmon?
00:41:08linuxstbNo, I don't even know what it is...
00:41:37Zagorlinuxstb: see Documentation/usb/usbmon.txt in the linux source
00:41:54Zagorit's basically tcpdump for usb
00:42:12Zagoralbeit a lot more confusing :)
00:43:26Zagorit's also what I've been staring at for the past week...
00:44:41linuxstbWill that help me if I can't control the USB device though?
00:44:46preglowman, this double-make-to-actually-link-the-codec bug is cool
00:44:55preglowcould we please have makefile expert fix this? :/
00:45:02Zagorlinuxstb: it might help you understand what's going wrong
00:45:21 Quit Nico_P (Remote closed the connection)
00:45:40Zagorit shows you all the low-level communication that is done to set up a usb device. it sounds like something there goes wrong.
00:45:51 Join iamben [0] (
00:46:19linuxstbZagor: Thanks, I'll try it, but can't at the moment as me and my device are away from a Linux machine.
00:47:07linuxstbpreglow: Doesn't the sim do the opposite - i.e. always relink codecs, even when they don't change?
00:47:35stripwaxrasher - maybe I'm missing something, what's the reason the backdrop setting wouldn't persist?
00:48:24preglowlinuxstb: i think maybe it does, i can't remember having the problem there
00:49:02 Quit ender` (" Debian comes in three flavours: stale, rusting and broken.")
00:49:26*linuxstb thinks Ender` was unfair...
00:49:43rasherstripwax: haven't looked at the code, but it's pretty much "just because" these days. Before moving to config files, it was to save space in the config sector.
00:50:42 Join J3TC- [0] (
00:51:06linuxstbZagor: BTW, congratulations on your USB progress. Do I understand correctly you've now got both read and write bulk transfers working?
00:52:06Zagorlinuxstb: yeah, with a few remaining issues
00:52:12stripwaxrasher - wow, that's pretty grim
00:52:59Zagorrasher: I agree not saving the backdrop is silly
00:54:22 Quit MethoS- (Remote closed the connection)
00:55:08 Join MethoS- [0] (
00:55:31preglowman, i would kill for jtag
00:56:01rasherpreglow: ask bagder to send the sansa
00:56:06rasherFinally put it to good use..
00:56:26 Join webguest54 [0] (i=477838fd@gateway/web/cgi-irc/
00:57:46scorche|was for the killing...well...we shall have a vote :)
00:58:07 Join Benoitb [0] (n=Benoitb@
00:59:44 Part Benoitb ("Kopete 0.12.5 :")
01:00:25Zagorwrite: 1048576 bytes in 560 ms = 1872 kB/s
01:00:38Zagorstill without optimisation
01:00:48Zagorthat's just using bigger packets
01:01:23linuxstbI was about to ask that - what size packets are you using?
01:02:02Zagorthis last was 512
01:02:12Zagorthat's the largest allowed
01:02:35LloreanDo we know what the OF uses?
01:02:52ZagorLlorean: all high-speed devices use 512
01:03:20 Part safetydan
01:03:21LloreanAh, good to know. Thanks.
01:03:29linuxstbWhen using dd with USB, I find a 32KB buffer size gives big speed improvements, so I assumed it was using that size over usb, but obviously not...
01:05:34rasherZagor: not bad
01:06:30LloreanI bet it's faster than disk mode on my Nano
01:06:32preglowrasher: it's not coldfire :/
01:06:55 Quit linuxstb (Read error: 104 (Connection reset by peer))
01:07:18 Join linuxstb [0] (
01:08:18J3TC-Hrmm...which bmp should I change for the iaudio X5 bootlogo?
01:08:31LloreanThe one that's 160 by something
01:09:16J3TC-rockboxlogo.160x50x16.bmp ?
01:09:26LloreanProbably, yes.
01:09:28linuxstbJ3TC: Check apps/bitmaps/native/SOURCES for the rules
01:09:34preglowamiconn: that 'illegal' trick is pretty nice
01:09:34Lloreanx2 means "2 bit color", so no.
01:09:36J3TC-Ah ok
01:14:19 Quit scorche|w ("CGI:IRC")
01:15:31***Saving seen data "./dancer.seen"
01:17:20 Quit stripwax ("Miranda IM! Smaller, Faster, Easier.")
01:25:51amiconntest_codec is still shaky
01:26:08 Quit MethoS- (Read error: 110 (Connection timed out))
01:26:29amiconnPlugging power while test_codec is running reproducably causes a prefetch abort at 0xDEADBEEE
01:26:31 Quit atsea-22 (Read error: 104 (Connection reset by peer))
01:26:33amiconn..on my H10
01:27:52 Join Shaid [0] (
01:28:11 Nick fxb is now known as fxb__ (
01:28:24linuxstbamiconn: Could that cause the codec thread to wake up?
01:28:53amiconnPlugging power broadcasts a SYS_ event
01:29:05linuxstbHmm, so mpegplayer would suffer similarly...
01:30:31 Join barrywardell [0] (
01:31:00amiconnHmm, weren't there reports that mpegplayer crashes when plugging usb?
01:31:07 Quit jhMikeS (Read error: 104 (Connection reset by peer))
01:33:28 Join jhMikeS [0] (n=jethead7@rockbox/developer/jhMikeS)
01:34:24linuxstbI don't know.
01:35:44jhMikeSyes, it will wake up the codec thread. why it needs to receive sys events I've no idea. I didn't check that simply because I wrongfully assumed it didn't. :\
01:36:11amiconnIf it has a registered queue, it will receive all broadcast events
01:36:43jhMikeSI know that...but _why_? it has no purpose I can see.
01:38:58amiconnI don't know. Does the codec thread need to do something on usb connection? I.e. discard some data instead of writing it back?
01:39:57amiconnIf not, the codec queue wouldn't need to be registered, and the SYS_USB_CONNECTED handling could be removed from the codec thread
01:40:09jhMikeSplayback should be stopped by the audio thread before the audio thread acks
01:40:59jhMikeSvoice will also be returned to current. imo voice and codec should just be private. only the audio thread need handle that.
01:41:30amiconnlinuxstb: Optimising scalarproduct for arm7 gives some further speedup. Not as much as vector_add/vector_sub though
01:41:41amiconnNow please someone optimise the entropy decoder...
01:43:53barrywardellamiconn: what's the status on your investigations of the PP LCD bridge? Have you had a chance to look at it any more?
01:44:08amiconnOnly a little bit
01:44:12 Quit Sedgewick (Read error: 110 (Connection timed out))
01:44:42amiconnI.e. I did a dump of my small H10 and then compared the 4 dumps side by side for a while
01:44:47amiconnNo experimentation yet
01:45:04amiconnI also tried to find the LCD routines in the H10 OF, but no luck so far
01:45:31barrywardellgiven the incredible speed improvement possibility for large H10 and ipod color, it would be nice to know more
01:45:41amiconn(I want that both for gleaning some details about the bridge timing and the TL1771 init sequence)
01:48:32barrywardellsearching for 70008a00 in an objdump gives a couple of hits
01:48:38 Quit Thundercloud (Remote closed the connection)
01:49:11amiconnYes - and those places put this value into a struct with an address calculated by another function...
01:49:48 Quit Zagor ("Client exiting")
01:49:54barrywardellah, I see that makes things a lot harder to figure out. Have you tried looking at the bootloader dumps instead?
01:50:22*jhMikeS bits shark? =>
01:50:35linuxstbamiconn: Do you have any numbers for the scalarproduct opt?
01:51:16amiconn-c2000 is now 74.5% (was 73.5% with add/sub, and 66% with svn)
01:51:45linuxstbAs you said, small improvements add up...
01:51:48amiconn-c3000 48% (43%, 37%)
01:52:27amiconn-c4000 still running...
01:52:47linuxstbSo with optimisation of the entropy decoding and dual-core, it seems -c3000 might be playable on PP...
01:52:53amiconnNote that the full patch would slow down gigabeat
01:53:16amiconnJust add/sub would be neutral - so I will keep the C version for gigabeat
01:54:20pixelmagah... TiMiD broke metronome a bit (for all it seems) and disabled it for hwcodec. He's not around ATM and it's unfortunate that it is only discovered now... :\
01:55:53amiconnlinuxstb: I don't know... even if one core does only filtering...
01:57:00 Join hcs [0] (n=agashlin@rockbox/contributor/hcs)
01:57:22linuxstbCan you remember your estimate of how much time is spent doing the entropy decoding?
01:58:28linuxstbAh, for Coldfire and -c1000, 63% entropy, 37% predictor...
01:59:46amiconnThat was before asm predictor....
02:00:41linuxstbpixelma: I wonder how he got away with disabling it for hwcodec...
02:00:52amiconnNow it's 75% entropy, 25% predictor
02:01:35linuxstbpixelma: Hmm, seems it was hidden in this commit -
02:01:37amiconn-c3000 is well within reach on coldfire with an optimised entropy decoder. It's already 108% realtime in my local version
02:01:46pixelmafor swcodec it's not broken but you use it differently which got me confused
02:02:13pixelmalinuxstb: that's what I based my statement on
02:02:36linuxstbpixelma: Have you tried compiling it for hwcodec?
02:03:05pixelmano, not yet
02:04:14pixelmaI also have to figure out why it doesn't work correctly with tapping and "playing" on the c200, I suspect button action conflicts :|
02:04:40amiconnlinuxstb: -c4000 on PP: 17.6% (full) <- 14.9% (add/sub) <- 11% (svn)
02:08:25 Quit jhMikeS (Nick collision from services.)
02:08:31 Join jhMikeS [0] (n=jethead7@rockbox/developer/jhMikeS)
02:09:00pixelmaI can see the potential conflict for the c200 keymap - the "play" button used to play the beat back is also the "up" used to adjust the volume...
02:10:39*pixelma is reminded of the plugin button action mail she never got a response to
02:10:54linuxstbamiconn: Nice improvement for -c4000 over svn...
02:11:35amiconnDid you see the entropy vs. predictor figures 10 minutes ago?
02:11:51linuxstbpixelma: To be honest, I'm not that fond of the button actions anywhere - I found the old method much clearer (apart from the messy pre- handling).
02:12:35linuxstbamiconn: Yes, and I've been staring at the entropy decoder...
02:12:59 Join kkurbjun [0] (
02:14:18pixelmalinuxstb: I was also a bit lost yesterday when looking at the c200 keymap file, trying to understand what's happening in the radio screen. But I don't know the "old method" (except in plugins)
02:15:51linuxstbThe "old method" is just the same as the plugins - defining the buttons directly.
02:16:41 Quit webguest54 ("CGI:IRC (EOF)")
02:17:16amiconnThe old method is worse when the same, or very similar, button->action mapping is needed in many places: lots of duplicated code
02:17:37 Quit joshin (Read error: 104 (Connection reset by peer))
02:17:51amiconnSo imho the action system makes sense in the core. But for plugins, I like the old method better.
02:18:27 Quit jhMikeS (Nick collision from services.)
02:18:28LloreanI agree with amiconn on this one
02:18:33 Join jhMikeS [0] (n=jethead7@rockbox/developer/jhMikeS)
02:18:49 Join joshin [0] (n=joshin@VDSL-130-13-11-27.PHNX.QWEST.NET)
02:22:32 Join safetydan [0] (i=dc9d468b@rockbox/developer/safetydan)
02:23:05*linuxstb sleeps
02:23:06 Part linuxstb
02:23:25preglowi need a brain center for asm
02:23:46preglowthis isn't working out
02:25:58 Join sdoyon [0] (
02:28:39 Join atsea-22 [0] (i=atsea-@gateway/tor/x-15cbfff90f7cd4b1)
02:34:26 Join JdGordon [0] (
02:38:16 Quit jhMikeS (Read error: 104 (Connection reset by peer))
02:38:47 Join pianohacker [0] (
02:38:56 Join jhMikeS [0] (n=jethead7@rockbox/developer/jhMikeS)
02:39:04pianohackerWhat's the difference between apps/player/keyboard.c and apps/recorder/keyboard.c ?
02:39:25sdoyonWhat is the intended effect of NEXTREPEAT in the radio?
02:40:28 Quit sup ("Lost terminal")
02:40:30 Join Shaid` [0] (
02:40:38sdoyonpianohacker: AFAIK those are two different Archos models.
02:41:20pianohackerOkay. They both have kbd_input functions; which is the canonical one?
02:41:40JdGordonpianohacker: player is sued for the charcell target, recorder is used for the rest
02:41:53pianohackerOkay, thanks.
02:42:16pianohackerNow I can rewrite it.
02:42:21 Quit pianohacker (Client Quit)
02:44:55safetydanrewrite? this should be interesting
02:48:30sdoyonjhMikeS: would you glance at the last patch in P#6239 and give me a general comment?
02:54:01pixelmametronome even compiles for hwcodec and works but the keymap at least for the Ondio is pretty much unusable too :\
02:56:03jhMikeSsdoyon: quickly...then sleep time
02:57:59 Quit Shaid (Read error: 101 (Network is unreachable))
02:58:01 Nick Shaid` is now known as Shaid (
02:58:49jhMikeSsdoyon: I should mention on hwcodec the event isn't available but that _is_ interrupt context so can't be used there anyway. Not sure it matters.
03:01:38jhMikeSsdoyon: I don't think you need a manul reset event either. The callback will signal it, and the wait will leave it nonsignaled
03:01:59 Part pixelma
03:04:01sdoyonjhMikes: Oh? I'll go reread kernel.c I guess.
03:04:49sdoyonjhMikeS: about HWCODEC and the callback in interrupt context: what do I do about mutual exclusion?
03:05:11jhMikeSdisable interrupts and reset to old level
03:05:48jhMikeSusing auto event and just the event_wait call alone would perform atomically what it seems you're getting at
03:07:23sdoyonjhMikeS: there's that queue_write that both queue_clip and the callback manipulate, and conditions on queue_level == 0...
03:08:28sdoyonjhMikeS: the adding of the silence cip at the end from within the callback looks a bit problematic without a mutex...
03:09:12sdoyonjhMikeS: OK I'll look at it again, look into HWCODEC and explain things better... just wanted a general opinion as to I was going in a worthwhile direction with this.
03:10:10jhMikeSon HWCODEC set_irq_level would act like the mutex and not let it be entered during the disable
03:10:28sdoyonjhMikeS: what's with the #if CONFIG_CPU == SH7034
03:11:06jhMikeSbetter go with #if CONFIG_CODEC == SWCODEC
03:11:51 Join newbyx86 [0] (
03:12:31sdoyonjhMikeS: if it were my choice yes, but that #if CONFIG_CODEC != SWCODEC is already in talk_force_shutup(). Which cases does that cover? They disable DMA in there...
03:13:48jhMikeSThat's done for hwcodec since the callback is ISR context there. The callback is just thread context on SWCODEC.
03:14:28 Quit BigBambi (Read error: 113 (No route to host))
03:15:35***Saving seen data "./dancer.seen"
03:16:40sdoyonjhMikeS: but what's the difference between #if CONFIG_CODEC == SWCODEC vs #if CONFIG_CODEC == HWCODEC?
03:18:16jhMikeSThere's no HWCODEC #define
03:18:27 Quit barrywardell ()
03:18:27jhMikeSI guess they already protect on hwcodec
03:20:07jhMikeShwcodec currently refers to the MAS decoder on archos models. SWCODEC uses the CPU to decode audio.
03:20:19sdoyonjhMikeS: Oh right, but then is #if CONFIG_CODEC == SWCODEC supposed to be equivalent to #if CONFIG_CODEC != SWCODEC?
03:21:21jhMikeSCONFIG_CODEC == SWCODEC is eq. to CONFIG_CODEC != some_hardware_decoder (!HWCODEC)
03:22:06 Quit Mouser_X (Read error: 110 (Connection timed out))
03:23:05sdoyonjhMikeS: Yes I know that, I'd just like to know how to handle the #if CONFIG_CPU == SH7034 that I see in talk.c, and whether I'm to understand that this #ifdef is intended to cover all HWCODEC cases.
03:23:17 Join Spiritsoulx [0] (n=eyes_of_@
03:23:20 Part Spiritsoulx
03:23:50sdoyonjhMikeS: CHCR3 &= ~0x0001; /* disable the DMA (and therefore the interrupt also) */
03:24:33sdoyonjhMikeS: That must be specific to that particular CPU. What other HWCODEC cases does this leave out?
03:24:52jhMikeS#if CONFIG_CODE == SWCODEC, #elif CONFIG_CPU == SH7034, #endif
03:25:57sdoyonjhMikeS: Ah good. Thanks.
03:26:08jhMikeSI suppose myself, I'd use the most technically correct manner even if redundant atm
03:27:12sdoyonjhMikeS: I guess the idea is there might be other HWCODEC archs in future. Just wasn't obvious to me that it covered all cases.
03:27:45sdoyonjhMikeS: OK then, good night :-). Thanks.
03:36:13 Quit Toxicity999 (Read error: 104 (Connection reset by peer))
03:37:33 Join Toxicity999 [0] (n=bryan@unaffiliated/Toxicity999)
03:40:56 Quit kugel ("ChatZilla [Firefox]")
03:51:45 Join qweru [0] (
04:00:54sdoyonAnyone understands what NEXTREPEAT is intended to do in the radio screen?
04:00:54 Quit donutman25 (Read error: 104 (Connection reset by peer))
04:02:54 Join sarixe [0] (
04:21:00 Join Davide-NYC [0] (
04:21:26 Quit sarixe ("(EE) Failed to load "quit" module")
04:21:29 Join viperman3 [0] (i=42928f49@gateway/web/cgi-irc/
04:22:29Davide-NYCHi, anyone care to test something for me?
04:22:51*Davide-NYC about to file a bug report and want to make sure it's not just me.
04:23:03sdoyonWhat kind of something?
04:23:22Davide-NYCI believe that mp3_encoder is fairly broken on h1x0 players.
04:23:39Davide-NYCActually totally broken. It freezes the player requiring a pin reset
04:24:06Davide-NYCcan someone with a coldfire target attempt to encode a wav to mp3 for me?
04:24:33Davide-NYCIt works for gigabeat f20.
04:24:43Davide-NYCso it's not all targets.
04:24:53 Join Mouser_X [0] (n=someone@
04:25:43 Join RoC_MasterMind [0] (
04:25:55sdoyonDavide-NYC: ouch. Hmm can't help you with this one, not until I make the plugin accessible :-).
04:27:50Davide-NYCJust for the sake of it I ran it in the h1x0 simulator (in cygwin) and it worked fine.
04:28:24 Quit sdoyon ("ircII EPIC4-2.4 -- Are we there yet?")
04:28:57 Quit miepchen^schlaf (Read error: 110 (Connection timed out))
04:29:04 Join miepchen^schlaf [0] (
04:29:13 Quit JRoT (Read error: 110 (Connection timed out))
04:31:55Davide-NYCAnybody else with a coldfire targetout there willing to confirm this?
04:32:27 Quit kkurbjun ("Leaving.")
04:33:29 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
04:39:08 Join karashata [0] (
04:42:32Davide-NYCmp3_encoder crashes bug filed: FS #8019
04:42:32 Quit viperman3 ("CGI:IRC (EOF)")
04:43:13Davide-NYCWhat would be great is a transcoder. Whatever decoder to whatever encoder. That would be sweet.
04:52:16 Join DC1 [0] (
05:00:01 Join CiNc028 [0] (
05:00:14CiNc028i have a question or two
05:00:22CiNc028 can anyone help me. about rockbox
05:00:48CiNc028im trying to get my ipod photo to playback vidz
05:01:40Mouser_XI know how to play MPEG vides, but I don't know about any VID format.
05:02:05CiNc028well if its mpeg. its best for me.
05:02:11*karashata baps Mouser_X here too...
05:02:21karashataCiNc028: what seems to be the problem?
05:02:28CiNc028now is rockbox the only thing out righ tnow to do that
05:02:55CiNc028well my problem is stupid home dvd player broke on me. and now i wanna see vidz. other than my pc
05:02:56krazykitCiNc028, yes. read the wiki on how to do that
05:03:27CiNc028thats what i was just reading just now
05:04:24CiNc028okay ima go read all of the info.
05:04:32CiNc028thnk u mouser
05:04:43CiNc028and krazy
05:04:51scorcheplease use real english in here, as per the guidelines
05:06:40 Join donutman25 [0] (n=chatzill@
05:12:58 Join hannesd__ [0] (
05:15:38 Quit bb (Nick collision from services.)
05:15:39***Saving seen data "./dancer.seen"
05:15:43 Join bb_ [0] (n=bb@unaffiliated/bb)
05:17:54 Quit Soap__ (Read error: 104 (Connection reset by peer))
05:18:05 Join Soap__ [0] (
05:25:25 Quit J3TC- (".•«UPP»•.")
05:25:54 Join kubiix [0] (
05:28:23 Quit hannesd (Read error: 110 (Connection timed out))
05:28:24 Nick hannesd__ is now known as hannesd (
05:31:52CiNc028hey mouser can i ask u something
05:32:08CiNc028no wi just got done reding upto where it talks about the usb .
05:32:33CiNc028 so if i connect my ipod via usb i cant transfer files to it
05:32:42CiNc028 sorri. im typing so fast
05:33:49Davide-NYCI vote to start using either esperanto or lojban in here exclusively. (just kidding)
05:34:22karashataCiNc028: you can't transfer files in Rockbox, you'd either have to boot into the original iPod firmware, or UMS mode (where the iPod is seen as a hard drive instead of an iPod)
05:34:43karashataDavide-NYC: that'd just be cruel...
05:34:54CiNc028 ok i understand totally now
05:35:03karashatahope that helps you
05:35:03CiNc028tiem to keep reading
05:35:23CiNc028it does. cus i understand now lol
05:35:39scorche"cus" is not english
05:35:56karashataI don't know if/when they'll get USB working in Rockbox, I think they're working on it though
05:37:03Mouser_XOne thing I'd like to see (for all targets, but that's a different matter) is longer battery life. I realize why the iPod lacking though.
05:37:40*karashata gets nearly 11 hours average on 20 GB H10
05:37:46scorcheMouser_X: well, for all targets except the PP devices, we already get better battery life than the OF
05:37:57CiNc028excuse me scorche. "because"
05:38:16Mouser_XI'm considering doing the iPod battery mod for my Gigabeat...
05:38:49CiNc028is there any sites to buy replacement faceplates for the ipod?
05:39:00CiNc028any sites anyone here recomend?
05:39:21karashatacould always check eBay for busted iPods for parts...
05:41:46 Quit RoC_MasterMind ("Leaving")
05:42:23 Join sarixe [0] (
05:42:33CiNc028im wondering about how far moddin ipods will go. maybe one day one can use it to play psp games off of.
05:43:03Mouser_XiPod hardware isn't good enough (not even close.)
05:43:03CiNc028why do you say nope?
05:43:15CiNc028<~~ bubbl ehas been busted
05:43:25CiNc028friggin space bar
05:43:53Mouser_XThere's some audio codecs that the iPod has trouble playing (SPC being the notable one. It does run, but some people have complained about its accuracey).
05:44:18karashataCiNc028: the processor in an iPod isn't nearly powerfully enough to emulate the PSP hardware no matter how you cut it...
05:44:46karashatait's a DAP, it's meant to play music, mainly
05:44:47Mouser_XIt can *barely* emulate the Gameboy (I'm somewhat surprised it runs at all.)
05:45:29CiNc028well i can say its good enough for music
05:45:46CiNc028oh i do have one other question
05:45:52Mouser_XThe Gigabeat can emulate the SNES though, if one were written and optimised enough for it.
05:46:05karashataCiNc028: what's your question?
05:46:10CiNc028if im installing rockbox on my ipod should have all and any other windows programs closed?
05:46:22karashatashouldn't matter
05:46:30Mouser_XOnly for the stability of the operating system.
05:46:56Mouser_XIn other words, in your context, it doesn't really matter. However, if your system is "shakey" then you might want to.
05:47:15CiNc028i have always played it safe even with moddin the psp. i would restart and do what i need to do.
05:47:16karashataMouser_X: that really shouldn't be such an issue unless you're running Win98 or something...
05:47:29CiNc028i only run winxp pro
05:47:34*Mouser_X *is* running Win98.
05:47:53CiNc028i didnt knwo people still ran that os
05:48:41karashatapost... Win2000 maybe, you don't need to worry so much about shutting everything down when installing applications and such
05:48:52krazykitthough this is getting offtopic...
05:49:18CiNc028okay ill just install as is.
05:50:06 Join courtc [0] (
05:53:22CiNc028okay no whow do i install rockbox?
05:53:31krazykityou read the manual.
05:53:35Mouser_XRead the manual.
05:53:40Mouser_XDang. Beat me.
05:53:41 Quit kubiixx (Read error: 110 (Connection timed out))
05:55:18CiNc028well what i mean is do i have to format the ipod HD ?
05:55:37Mouser_XNo, you don't.
05:55:49Mouser_XIs your iPod FAT32?
05:56:10krazykithe's on windows, it is.
05:56:13CiNc028let me check
05:56:19 Join psycho_maniac [0] (
05:56:30Mouser_Xkrazykit: D'oh! You're right.
05:56:45Mouser_XCiNc028: Nope, no need to format. You're good to go.
05:57:46 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
05:57:54CiNc028ok no format good
06:13:20 Quit lazka (Remote closed the connection)
06:13:38 Quit DC1 ("If Obi-wan ain't home then I don't know what the fsck we're gonna do. I ain't got no other connections on Tattooine.")
06:15:00 Quit sarixe ("(EE) Failed to load "quit" module")
06:15:36CiNc028ok im runnign rockbox now:) yay
06:17:09psycho_maniacwhat player?
06:19:20CiNc028thnk u thnk u
06:19:33CiNc028i would clap but no one will here me but the baby
06:20:03CiNc028now to get out of rockbox and back into ipod crap. how do i do?
06:20:23Mouser_XIt's in the manual somewhere...
06:20:28Mouser_X(I remember reading it.)
06:20:38CiNc028me too
06:20:45CiNc028ill read again
06:25:41Davide-NYCAny devs want to help a feeble wretch add in target and build info to the test_codec logfile?
06:26:25Davide-NYCI've figure where to put the log_test() funtion call but have not clue how to insert the target and build info.
06:28:08Davide-NYC(no laughing) This is what I have right now: log_text("This is where we insert the target and build info.",true);
06:28:35Mouser_XI'd help, but you already know more than I do.
06:28:59*karashata could only wish he knew C
06:29:06 Join webguest22 [0] (i=43a4557f@gateway/web/cgi-irc/
06:29:48 Quit webguest22 (Client Quit)
06:30:09 Join webguest27 [0] (i=43a4557f@gateway/web/cgi-irc/
06:32:04 Quit animeloe (Nick collision from services.)
06:32:06 Join animeloe_ [0] (n=animeloe@unaffiliated/animeloe)
06:32:26 Nick animeloe_ is now known as animeloe (n=animeloe@unaffiliated/animeloe)
06:33:16safetydanDavide-NYC: doesn't the credits plugins have version information?
06:34:04Davide-NYCsafetydan: looking... thanks for the tip
06:36:03midkayyes, plus the "info -> rockbox info" screen does.
06:41:31Davide-NYCmidkay: what c file is the info screen?
06:42:16midkaynot sure.. screens.c *perhaps*, or dig into.. main_menu.c to see what it references?
06:42:51 Quit webguest27 ("CGI:IRC (EOF)")
06:43:14Davide-NYCcredits called show_logo() so I thought that wasn't what I wanted.
06:43:42Davide-NYCnot even sure where show_logo() is defined
06:43:49CiNc028hey thanks again mouser.
06:44:00CiNc028 ima go watch some vidz now
06:44:27psycho_maniachave fun
06:47:16safetydanDavide-NYC: grep is your friend
06:47:51Davide-NYCnever became friendly with regular expressions. Too old. Too steep.
06:49:19safetydanDavide-NYC, grep doesn't require regexps. All you really need is something like "grep -r show_logo rockbox-src/"
06:50:19Davide-NYCbut I should RTFgrepM, right?
06:50:34Davide-NYCI'm on it.
06:51:07Davide-NYCI think I'm after appsversion[], although I'm not sure about those square braces.
06:59:09 Quit safetydan ("CGI:IRC (Ping timeout)")
07:05:23 Quit CiNc028 (Read error: 104 (Connection reset by peer))
07:09:31 Join webguest69 [0] (i=423a97a1@gateway/web/cgi-irc/
07:09:49webguest69oh, yes, I'm no longer banned
07:10:05JdGordonwe can change that if you want?
07:10:16webguest69Not again please!
07:11:21 Nick webguest69 is now known as DavidRawson (i=423a97a1@gateway/web/cgi-irc/
07:12:23DavidRawsonDo you think that there will ever be a possiblity of a more complex rockbox system?
07:12:50Davide-NYCI'm making it more complex right now!
07:12:55DavidRawsonOne that can run separate binaries instead of plugins.
07:13:01Bagderit's more complex today than it was a week ago
07:13:44BagderDavidRawson: why would that be a good idea?
07:13:59ShaidI'm off to write code that randomly assigned button bindings so that each time you boot rockbox it's a guessing game which button does what.
07:14:09Shaidwatch out for when select gets bound to 'delete all'
07:14:16DavidRawsonMake the system run on binaries that don't have to be compiled with rockbox, that can run separately.
07:14:41Bagderand I repeat: why would that be a good idea?
07:15:02DavidRawsonTo make it easier to run programs on the system.
07:15:13Shaidwouldn't that make it harder?
07:15:34BagderI don't think people think running plugins is hard
07:15:42***Saving seen data "./dancer.seen"
07:15:47ShaidI mean, most of the platforms are on different cpus
07:16:02Davide-NYCwhat exactly does snprintf return?
07:16:05Shaidso you'd have to write a seperate binary type for each one, unless I'm being thick.
07:16:09 Quit jhulst (Remote closed the connection)
07:16:10DavidRawsonScripts that rockbox can run?
07:16:14BagderDavide-NYC: the length of the new string, iirc
07:16:44DavidRawsonLike an actionscript works for macromedia flash.
07:16:54Shaidthen they're not binaries.
07:17:11DavidRawsonYour different CPUs made me change my mind
07:17:19Davide-NYCBagder: so what exactly does this do? snprintf(version, sizeof(version), "Ver. %s", appsversion);
07:17:33scorcheDavidRawson: new IP i see...
07:17:52DavidRawsonno, I asked Lloren to unban me
07:17:53BagderDavide-NYC: it creates a string in a buffer with the version number
07:18:25Davide-NYCcan I embed that whole business inside a function that expects a string as an argument?
07:18:25DavidRawsonand how could I get a new IP anyway?
07:18:42 Join [omni]_ [0] (
07:19:01Davide-NYCBagder: example: log_text("snprintf(version, sizeof(version), "Ver. %s", appsversion);",true);
07:19:30Davide-NYCmaybe there's an extra ";" in there'
07:19:49 Join bb [0] (n=bb@unaffiliated/bb)
07:19:52DavidRawsonnone of that stuff makes sense to me
07:20:04 Join andrewg867 [0] (
07:20:19DavidRawsonif this is a coding IRC, i'm outta here!
07:20:25 Quit DavidRawson ("CGI:IRC (EOF)")
07:20:26*Davide-NYC throws caution to the wind and compiles
07:20:37 Quit shodanX (Remote closed the connection)
07:20:40 Join shodanX [0] (
07:20:42*scorche wonders who unbanned him..
07:22:51 Join atsea-32 [0] (i=atsea-@gateway/tor/x-248d0bfa7dfc743e)
07:22:55Davide-NYCif I just call appversion what do I get?
07:23:58scorchenope...he is still is a new IP
07:25:02Davide-NYCsomeone please take mercy on me and walk me through this valley of darkness.
07:25:23 Join animeloe_ [0] (n=animeloe@unaffiliated/animeloe)
07:31:12 Quit jeremyb (Connection timed out)
07:31:26 Quit Toki_ (Read error: 110 (Connection timed out))
07:31:33 Join ackbahr [0] (
07:31:35 Quit andrewg877 (Read error: 110 (Connection timed out))
07:32:14 Quit delYsid (Read error: 110 (Connection timed out))
07:32:16 Quit bb_ (Read error: 110 (Connection timed out))
07:32:50 Quit alienbiker99 (Read error: 110 (Connection timed out))
07:33:08 Quit [omni] (Read error: 110 (Connection timed out))
07:33:58 Join delYsid [0] (
07:35:27 Quit RaRe (Read error: 110 (Connection timed out))
07:35:55 Quit animeloe (Read error: 110 (Connection timed out))
07:36:45 Join davina [0] (
07:36:51 Join alienbiker99 [0] (
07:37:08 Quit psycho_maniac (" HydraIRC -> <- The alternative IRC client")
07:37:25 Join jeremyb_ [0] (n=jeremy@unaffiliated/jeremyb)
07:38:54 Quit jeremyb_ (Client Quit)
07:39:04 Quit atsea-22 (Read error: 104 (Connection reset by peer))
07:39:05 Join jeremyb [0] (n=jeremy@unaffiliated/jeremyb)
07:39:40 Part jeremyb
07:40:06 Join ompaul [0] (n=ompaul@gnewsense/friend/ompaul)
07:42:58 Join Nico_P [0] (n=nicolas@rockbox/developer/NicoP)
07:44:44 Join ddalton [0] (
07:45:09ddaltonhow could I reset the alarm to go off once it has gone off already?
07:45:13ddaltonin code?
07:45:55ddaltoncould i change this to true? rtc_enable_alarm(false);
07:49:32 Nick animeloe_ is now known as animeloe (n=animeloe@unaffiliated/animeloe)
07:51:03 Quit Nico_P (Remote closed the connection)
07:54:24 Quit joshin (Read error: 104 (Connection reset by peer))
07:54:27 Join joshin [0] (n=joshin@VDSL-130-13-11-27.PHNX.QWEST.NET)
07:55:06 Join n1s [0] (
08:05:28 Join CiNc028 [0] (
08:09:50Davide-NYCI give up.
08:09:58Davide-NYCFor now.
08:10:05 Quit Davide-NYC ("ChatZilla [Firefox]")
08:10:06 Join CaptainSquid [0] (
08:11:32 Quit ompaul (Client Quit)
08:22:48 Join RaRe [0] (
08:35:19ddaltonhow do I use prep?
08:35:24ddaltonor what ever it is called?
08:37:14 Join Rob2222 [0] (
08:41:38GodEater_do you mean grep ?
08:44:04 Join Toki [0] (
08:45:34 Join animeloe_ [0] (n=animeloe@unaffiliated/animeloe)
08:55:40 Quit Rob222241 (Read error: 110 (Connection timed out))
08:55:50 Join grndslm [0] (
08:57:19scorcheddalton: there are an infinite amount of grep tutorials on the internet...
08:59:01GodEater_but the quick answer is : grep "stuff you're looking for" files_to_search
08:59:08 Quit animeloe (Read error: 110 (Connection timed out))
09:00:27 Join ender` [0] (
09:03:31 Join homielowe [0] (
09:10:08 Quit CaptainSquid ("Miranda IM!")
09:15:22 Quit animeloe_ ("Leaving")
09:15:45***Saving seen data "./dancer.seen"
09:15:52 Join CaptainSquid [0] (
09:16:54 Join petur [0] (n=petur@rockbox/developer/petur)
09:17:52 Join Zagor [0] (n=bjorn@rockbox/developer/Zagor)
09:19:08 Join animeloe [0] (n=animeloe@unaffiliated/animeloe)
09:21:16 Join leftright [0] (i=d9e1f960@gateway/web/cgi-irc/
09:23:16leftrightHi, H140, r15282: is it me or is the "play next" insertion broken ?
09:24:26leftrightnever mind, just not awake
09:26:39leftrightit does work as advertised
09:26:56 Quit joshin (Read error: 104 (Connection reset by peer))
09:27:03Mouser_X It's advertised?
09:27:33leftrightmeaning it works as intended
09:27:58Mouser_XAh. That's different. (Perhaps I've been up *way* too late...)
09:29:46 Join joshin [0] (n=joshin@VDSL-130-13-11-27.PHNX.QWEST.NET)
09:31:19 Join LinusN [0] (i=linus@rockbox/developer/LinusN)
09:48:14 Quit crashd ("leaving")
09:48:20 Join crashd [0] (
09:51:16 Join thegeek [0] ("> <−− linuxstb's new development toy :)
10:00:53 Quit hcs ("Leaving.")
10:01:17ZagorGodEater: any fun hw inside?
10:01:24Zagorapart from the dab receiver, I mean
10:03:29 Join pondlife [0] (n=Steve@rockbox/developer/pondlife)
10:04:12Zagorah, which one?
10:04:17GodEater_one sec
10:04:26BagderTCC773L wasn't it?
10:05:02GodEater_the switch is connected to the NAND and USB points just next to the TCC773L
10:05:13GodEater_to allow it to boot from NAND or USB
10:05:22GodEater_(at least, that's the theory)
10:05:33Bagderusb boot is really neat
10:05:42GodEater_it will be our first DAB target I believe :)
10:06:10Zagor"Wire wound", is that when you cut yourself on a sharp wire? :)
10:06:10*GodEater_ isn't sure how booting over usb is supposed to work - but linuxstb seemed confident
10:06:27GodEater_Zagor: hehehe
10:06:30 Join pixelma [0] (i=pixelma@rockbox/staff/pixelma)
10:06:33BagderGodEater: I'm quite sure we have a good idea on how that works... :-P
10:06:53 Part leftright
10:07:00GodEater_Bagder: glad to hear it
10:08:16 Quit thegeek_ (Read error: 110 (Connection timed out))
10:12:02BagderGodEater_: seeing my PM?
10:14:55 Quit ackbahr (Connection reset by peer)
10:15:48GodEater_I love the way the thing *advertises* it does DRM on the bloody case
10:15:59GodEater_like it's something to be proud of
10:16:00ZagorGodEater: yeah I noticed that too
10:17:04 Join pepie34 [0] (
10:20:30ddaltonGodEater_: what do I run that from? emacs? or the terminal?
10:20:49ddaltonre gprep sorry
10:21:01ddaltonhis not here?
10:21:58Bagderddalton: grep can be done in the terminal or within emacs, at your choice
10:22:33ddaltonGodEater_: so gprep SOUND_VOLUME trunk? I want to search for sound_volume
10:23:12 Quit CaptainSquid (Remote closed the connection)
10:24:02 Join CaptainSquid [0] (
10:24:09GodEater_ddalton: "grep -R SOUND_VOLUME ."
10:25:16 Join joshin_ [0] (n=joshin@VDSL-130-13-11-27.PHNX.QWEST.NET)
10:25:16 Quit joshin (Connection reset by peer)
10:28:13Zagorwhat card type does the ondio take?
10:29:22Zagorok, thanks
10:33:02 Nick fxb__ is now known as fxb (
10:36:45ddaltonok thanks let me try...
10:37:13ddaltonwhat should happen now?
10:39:36ddaltonGodEater_: what should I see happen in the terminal?
10:39:45ddaltonor hear :-)
10:40:12GodEater_ddalton: one second...
10:40:13ddaltonok maybe I typed it wrong
10:40:26ddaltonoh tnaks
10:41:16GodEater_ddalton: ah my mistake
10:41:25GodEater_ddalton: replace the "." with a "*"
10:41:36GodEater_putting "." there makes it scan your entire disc - not a good plan
10:41:51GodEater_ctrl-c it if it's still running
10:50:15amiconnGodEater: ?? "." is current dir ...
10:50:20 Quit TentaclePorn ()
10:50:26GodEater_amiconn: which contains ".."
10:51:09amiconn* would also cover that
10:51:18rasher* doesn't cover dotfiles
10:51:26*GodEater_ was just going to say that
10:51:30amiconn. also shouldn't
10:51:39GodEater_it does though I'm afraid
10:52:00GodEater_tip: don't ever "chmod -R ." - you *will* break your machine
10:52:51amiconn. doesn't include ..
10:53:27 Quit pepie34 ("Ex-Chat")
10:53:34GodEater_amiconn: just try it and see what happens
10:53:39amiconnI did
10:54:03 Join Thundercloud [0] (
10:54:05amiconn(for grep)
10:54:08GodEater_perhaps cygwin's grep works differently then
10:54:18amiconnI tried both cygwin and linux
10:54:26GodEater_but here, for me, "grep -R stuff ." goes wandering off all over the place
10:55:29amiconncygwin grep *is* gnu grep
10:55:40amiconnThen you might have some alias for it
10:56:07GodEater_or you might have a GREP_OPTIONS set
10:56:49GodEater_well I can't explain it then. Because it's definitely going back up the tree here
10:57:57 Join barrywardell [0] (
10:58:00GodEater_and clearly is doing that for ddalton too - since he's had no output
10:59:38 Join Arathis [0] (
11:00:03amiconngrep -V ==> grep (GNU grep) 2.5.1 (cygwin), GNU grep 2.5.3 (linux)
11:00:21 Quit RaRe (Read error: 104 (Connection reset by peer))
11:00:27rasherDoesn't here
11:00:50 Join RaRe [0] (
11:01:12n1sdoesn't here either...
11:01:20pixelmaI get the same amount of hits when doing the both versions - I searched for a Rockbox specific term in one of my trees. If it included ".." then it should have found twice as much because I got a second tree isn't imagining it either
11:01:54GodEater_I just ran lsof on the process
11:02:01GodEater_and it's off somewhere in my /usr/lib tree
11:02:29GodEater_ah no - hang on
11:02:32GodEater_I know what it's doing
11:02:48GodEater_ignore me - go about your business
11:02:49 Join lee-qid [0] (
11:02:57amiconnI can only imagine one of 3 things: (1) you have a buggy grep. (2) you have some option set that makes it include dotfiles (3) It depends on the filesystem
11:03:20GodEater_amiconn: it *is* including dotfiles, but not ..
11:03:27GodEater_it's searching through the .git directory
11:03:31GodEater_which is enormous
11:04:01amiconnSo why does it go up for you?
11:06:52GodEater_it doesn't
11:07:05ddaltonHmmm anyone know where Sound_volume is declared?
11:07:05GodEater_the lsof is just listing the libraries that grep is linked against.
11:07:07GodEater_<−− embarassed
11:08:44pondlifeGodEater_: Changing subject quickly - how's MoB workiing for you?
11:09:09GodEater_pondlife: fine currently - but haven't updated since whenever we were last speaking and I was complaining about crossfade buggering everything up
11:09:23pondlifeOK, so no new issues.
11:09:31GodEater_has Nico done anything much with it since then ?
11:09:53pondlifeLooks like it
11:10:00pondlifeI've just updated my H340
11:10:16*GodEater_ git pulls
11:10:29pondlifeI'm not convinced about his new buffer-when-disk-spinning strategy though
11:10:43pondlifeSuspect it could be simpler to use the callback
11:11:18pondlifeI might be misunderstanding his comments, but it seems like he no longer fills the buffer if the user spins up the disk.
11:11:30pondlifeWill play.
11:12:38*GodEater_ runs make
11:12:53*ddalton Looks for "Sound_volume"
11:13:50pondlifeddalton: Are you being case-insensitive?
11:13:57ddaltonJdGordon: do you know where SOUND_VOLUME is declared or defined?
11:14:00ddaltonwhat file
11:14:29ddaltonpondlife: it is upper case
11:14:39pondlifeYep, just checking
11:14:42*ddalton Tries grep once more
11:15:01ddaltonbut my find in my text editor doesn't check for case...
11:15:25GodEater_it looks to me like it's defined in the drivers for the different DACS
11:15:49***Saving seen data "./dancer.seen"
11:16:18GodEater_ddalton: so have a look in the files under firmware/drivers/audio
11:16:59*ddalton takes GodEater_'s advice
11:17:44ddaltonGodEater_: what's the h300? how do I find out?
11:18:15GodEater_ask pondlife :)
11:18:56*pondlife looks
11:18:57GodEater_ddalton: or check the wiki - the components should be detailed there
11:19:22 Join pepie34 [0] (
11:19:38GodEater_pondlife: running a shiny new MoB build
11:19:40GodEater_will keep you posted
11:19:56 Quit pepie34 (Read error: 104 (Connection reset by peer))
11:19:58GodEater_although I'm off to chilly Canada tomorrow - and have no idea how much intertubes access I'll have
11:20:22pondlifeKeep a notepad handy
11:20:50 Join mirak [0] (i=c34458b5@gateway/web/cgi-irc/
11:21:49pondlifeddalton: uda1380.c
11:22:13ddaltonpondlife: how did you find out that?
11:22:14pondlifeI should know that - the UDA went pop in my H340 - LinusN fixed it
11:22:34pondlifeBut, you can check config-*.h for details of any device
11:22:52pondlifee.g. config-h300.h here
11:23:20pondlifeMaybe the #define HAVE_UDA1380 could be commented better :)
11:23:53pondlifeddalton: Why do you need to know such low-level stuff, btw?
11:24:41ddaltonPondlife: I am making a new setting for the sound settings. It is "Alarm volume" so that is how loud the alarm will be when it wakes up...
11:25:08ddaltonWon't put it on the tracker because the coding is shocking but when I start coding proply again I will fix it.
11:25:19ddaltonbecause im probably going to stop after this.
11:25:20pondlifeI wouldn't think that needs to be low-level
11:25:25ddaltonfor a while maybe 5 months
11:25:40pondlifeLook at the "beep volume" setting
11:25:48ddaltonit seems to be. That is where it was declared...
11:25:48 Quit mirak ("CGI:IRC (Ping timeout)")
11:26:10 Join mirak [0] (
11:26:20ddaltonI was looking at volume. But might as well see what happens. I will do tht next time.
11:26:41 Quit Isolinear (Read error: 110 (Connection timed out))
11:27:00mirakI have a problem encoding videos on Ubuntu since edgy. the sound is broken when encoding with mencoder using mp3
11:27:11mirakthere is only white noise for sound
11:27:34GodEater_mirak: that's not a rockbox problem...
11:27:42ddaltonPondlife: what do you think of these tasks? FS #8006, 8007 and 8012?
11:27:58ddaltonobviously 8012 needs a bit of work...
11:28:03mirakGodEater for mpegplayer that's a problem, since the lines to encode are on the wiki
11:28:29Lloreanmirak: Just because the lines to encode are in the wiki doesn't mean it's our job to make sure they work on every OS or distro.
11:28:38pondlifeI like the idea of 8006, but wouldn't make it configurable.
11:28:52LloreanFor example, anything using ffmpeg on Ubuntu is unlikely to work if you've downloaded ffmpeg with apt-get, because they don't include MP3 support.
11:29:04pondlifeJust use the menus setting?
11:29:13pondlifei.e. talk_menu
11:30:09 Quit idnar (Nick collision from services.)
11:30:12 Join idnar_ [0] (n=mithrand@unaffiliated/idnar)
11:30:14pondlifeddalton: 8007 I think should be incorporated into a keyclick option (I started one a while back)
11:30:34ddaltonpondlife: what do you mean exactly?
11:30:47pondlife8012 is good, it should speak everything as displayed
11:31:32ddaltonpondlife: do you know what the blanks are? ( the answer to my comment?)
11:33:41pondlifeInfo screen is: version, buffer size, battery percentage and estimated runtime, disk capacity and disk free space
11:34:18pondlifeThe voicing should be identical, in the same order
11:34:47mirakLlorean so you see I get the answer to my question
11:34:51pondlifeIs the time voiced too? That's on the status bar for sighted users (so could go first).
11:35:23mirakLlorean thank you
11:35:42mirakLlorean I know where to find a fix now
11:37:00pondlifeGodEater_: Opinion, please; say your audio buffer is about a third full and the user does something that spins up the disk. Should the buffering thread take the opportunity to fill the buffer? Or should it wait until the buffer is "low" ?
11:37:13GodEater_it should fill
11:37:26pondlifeOK, will file a bug
11:37:42pondlifeIs loading a plugin enough to spin up the disk (I assume so)?
11:38:08LloreanFor 700mb of music, you've got the same total "buffering" time no matter how you divide it up pretty much, but if you can do anything to reduce the number of spinups exclusively for buffering, that should increase battery life a little I'd think.
11:38:19LloreanEspecially since I hear the archoses use a ridiculous amount of power to spinup
11:38:20ddaltonPondlife: I am just working on a better version. Give me a minute. Then if it is ok can you commit? Blank options isn't good enough and blind users are probably confused.
11:38:35pondlifeddalton: What is wrong with SVN?
11:39:40pixelmapondlife: my Archos doesn't :P
11:39:59pondlifepixelma: Nor does any target with SVN AFAIK :)
11:40:39 Join kugel [0] (i=kugel@unaffiliated/kugel)
11:40:57 Quit barrywardell ()
11:40:58pixelmaI meant using a lot power to spin up the disk...
11:41:21pondlifeAh, maybe flash needs a different strategy?
11:41:37LloreanFlash, I think, doesn't need a different strategy.
11:41:41pondlifeActually there's no reason not to rebuffer
11:42:31LloreanI remember someone having a ridiculously small buffer on Nano and claiming it didn't show any significant change in battery life.
11:44:28ddaltonpondlife: what about svn? I didn't say anything about it did I?
11:44:49pondlifeddalton: I meant, what does the patch actually change?
11:45:02pondlifei.e. What is currently missing?
11:45:07amiconnOn flash it would pay off to minimize the amount of data read and written instead of the number of spinups
11:45:45pondlifeSo only read when buffer low...
11:45:48Lloreanamiconn: In terms of buffering isn't the total amount of data read/written fixed? Or are you talking about only buffering the minimum each time (this song, next song)?
11:45:54amiconnBut for playback, the amount of data read doesn't change with rebuffering strategy, unless someone causes additonal disk accesses by loading plugins *and* skips back afterwards
11:46:26amiconnLlorean: It's fixed as long as special conditions don't cause extra reloads
11:46:34rasherconsidering that my sansa screeches when buffering, I'd like if that happened as rarely as possible.
11:46:45ddaltonPondlife: half the items don't speak.
11:46:46amiconnSo for simplicity flash targets should use the same strategy as hdd targets
11:47:08pondlifeddalton: Well, they all should. So the patch would be welcome
11:47:27LloreanCan we read more slowly on Sansa?
11:48:48 Join AlexC [0] (
11:49:47pixelmarasher: your Sansa screeches when accessing the flash? Mine doesn't but I have a the impression that the overall noise level is higher in Rockbox compared to the OF - hard to ptoof because of the huge difference in volume settings though. Maybe something isn't completely done right yet?
11:50:18rasherpixelma: no idea. I haven't used the of more than a few seconds
11:50:24rasherI could check
11:50:37Lloreanpixelma: I'd have to agree.
11:50:52LloreanI'm *almost* sure there's more buffering noise in Rockbox.
11:51:04 Nick idnar_ is now known as idnar (n=mithrand@unaffiliated/idnar)
11:51:13LloreanBut it may be that the minimum volume in the OF is higher than I normally listen to.
11:51:39amiconnSome measurements would probably tell
11:51:41pixelmadefinitely the same here
11:52:07pixelmaLlorean: I can't hear the buffering but there is already more noise if it just stays in the menu
11:52:19Lloreanpixelma: Do you have the wheel light enabled?
11:52:46ddaltonpondlife: well no point in my opinion in voicing as the same as the sighted users see. Otherwise we will hear version first. That is not what is wanted. So just speak it in the order I think is good: battery level, time, disk info, multy volume, version and buffer size
11:52:54ddaltonpondlife: what do you think?
11:53:01Lloreanddalton: Voice is not just for blind users.
11:53:02rasherAn mp3 of silence should allow to test this.
11:53:12pixelmaLlorean: well it's a button light (c200) but yes, I have it enabled with a short timeout. But it's on the OF too, coupled with the backlight
11:53:33ddaltonI know but you get what im saying. You don't have to speak exactly what's on the screen. It is just a list. As long as you find out the info
11:53:36Lloreanpixelma: For me at least, the OF showed about as much noise as Rockbox when LCD + button light are on.
11:53:41ddaltonnot a menu
11:53:44pondlifeddalton: The order should be the same. Time, version, buffer, ...
11:53:56Lloreanddalton: You should speak things in the order on screen. If you dislike the order, suggest the order change, as you said, it's just a list.
11:53:56pondlifeIt's confusing for support otherwise.
11:54:05pixelmaLlorean: maybe there are differences between e- and c-series
11:54:07 Join pepie34 [0] (
11:54:10Lloreanpixelma: Very possible.
11:54:37ddaltonpondlife: we aren't interested in the version. I check it once every 3 months or something. I check the battery level a lot and the time. Not so much the disk free info but a bit. then the version and buffer size are sort of not important.
11:55:05pondlifeddalton: Your patch should only add voicing to the existing screen (in the existing order). A second patch could then change the sequence (for screen and voice).
11:55:35ddaltonPondlife: that's what it does. But it isn't good. You hear version first...
11:55:55pondlifeThat's what the screen says first
11:56:00 Nick fxb is now known as fxb__ (
11:56:13pixelmaddalton: the version is interesting and important if you want to report a bug for example
11:56:15pondlifeAlthough I suppose battery level and time are on the status bar, so they could be put first
11:56:17ddaltonwhy speak the exact same item that is on the screen? in this case? you hit select it does the same thing on anything
11:56:26Lloreanpondlife: In that case they'd have to be put twice.
11:56:50pondlifeRight. So time, version, buffer, battery, disk
11:57:10Lloreanpondlife: Does that screen display the time separate from the status bar?
11:57:21 Part mirak
11:57:23ddaltonpixelma: come on. How offen do you check that. Yes it is a bit inportant but battery level and stuff should go before it.
11:57:27LloreanHonestly, I'm going to have to say I don't think objects from the status bar should be voiced in that screen.
11:57:38LloreanIf the status bar needs voiced, that should be something separate.
11:58:07ddaltonstuff like the time and battery level are already voiced in that screen...
11:58:35pondlifeLlorean: Time already got added (otherwise I agree with you)
11:58:36pixelmaddalton: I check that more often for version than for battery level (but I'm aware that this is different to you because the battery level is also shown in the statusbar...)
11:58:41Lloreanddalton: Things should be voiced as they're displayed. It creates confusion otherwise, for people who can see and just for general use. What's wrong with simply voicing it as displayed, and then putting in a feature request to re-order the screen?
11:58:54pondlifeddalton: The point is the ordering is a *seperate* issue.
11:59:02ddaltonpondlife: hang on
11:59:11*pondlife hangs on
11:59:21Lloreanpondlife: As in, SVN already voices time on that screen?
11:59:34LloreanThis is problematic, I think.
11:59:35pondlifeWe could remove it.... ;p
11:59:48*ddalton Must have started another argument
11:59:53*ddalton He is good at that
11:59:54LloreanI really, really don't like the idea of it voicing things that aren't even necessarily on-screen
12:00:02pondlifeI think we need voiced plugins firstr
12:00:04LloreanDoes it at least not voice time if the status bar is disabled?
12:00:09pondlifeSo clock can be voiced
12:00:15pondlifeNot sure, willhcheck
12:00:21pondlifeAh, typing failure
12:01:06pondlifeI'm afraid it reads the time regardless :(
12:01:12LloreanAh, even worse.
12:01:17n1sLlorean: I don't think time is voiced anywhere else so removing it would be bad
12:01:18ddaltonhalf a comment from sdoyon: "have to just ignore what's on the screen and put stuff we want in the order we want."
12:01:27pondlifeBest leave it in for now, then when clock is voiced we can sort it out
12:01:29ddaltonfor the info list in a email he sent me
12:01:31Lloreann1s: The solution then is "create an appropriate place for time to be voiced"
12:01:41rasherpondlife, Llorean: I think it'd make sense to add the clock to the Rockbox info screen anyway
12:01:41Lloreann1s: Even if it's as minor as actually placing a visible display of time on that screen.
12:01:44pondlifeddalton: sdoyan is wrong
12:01:47Lloreanrasher: No problem with that
12:01:51ddaltonsdoyon knows what he is talking about...
12:01:53n1sLlorean: i guess that's what that screen is to most people
12:01:59Lloreanrasher: My problem is with voice not matching screen, not the choice of what is displayed.
12:02:04rasherLlorean: I agree.
12:02:07ddaltonpondlife: your wrong :-)
12:02:22rasherI just think the solution should be to add a clock on the screen - not remove the voicing of time
12:02:22Lloreanddalton: No, he is.
12:02:26pondlifeThe UI is the UI, whether it's seen or voiced
12:02:37rasherSince some users might not have a clock visible anywhere else
12:02:41ddaltonwhat's the problem? where not saying voice 16 db in the sound settings screen when it is on 32 db for example
12:02:59Lloreanddalton: The problem is that the User Interface should be consistent.
12:03:02*JdGordon tihnks the time screen should be more accessable and voie it there, not the info screen
12:03:19Lloreanddalton: There should never be a case where voice speaks something that isn't there, or speaks things in an order that is different from the screen.
12:03:28Lloreanddalton: If the order on the screen is bad, FIX THE ORDER, don't just voice it differently.
12:03:32amiconnJdGordon: What time screen?
12:03:38rasherJdGordon: having a central location for "various sorts of info" is convenient though
12:03:45JdGordonexacrtly.... we dont have one other than the set time/date screen
12:03:46rashereven if there's a slight duplication going on
12:03:46pondlifeMaybe time should be on the info screen?
12:03:56rasherpondlife: I think it should
12:04:03JdGordonrasher: it shouldnt be named rockbox info then.. maybe just info?
12:04:16rasherDon't see why
12:04:24rasherThe battery isn't part of rockbox either
12:04:27pixelmaddalton: and I use voice often even though I'm not blind if voice speeks something else than I can see...
12:04:28pondlifeDisk space left isn't really rockbox info
12:04:39pondlifeI use voice whenever I'm driving
12:04:39rasherPretty much only version is really rockbox info
12:04:52pondlifeI want the UI to be as I remembered it
12:05:00JdGordonok, then we add time/date to the info screen?
12:05:23JdGordonthe yeps have it!
12:05:27pondlifeIt makes sense to put it first too (if noone objects)
12:05:57JdGordonyeah, time, date, svn build, gap, battery, gap, disk space
12:06:31n1sThe only thing about that screen that annoys me is when I accidentally push the select button and it starts scanning the disk, urgh
12:06:37pondlifeJdGordon: Why does the info list have a selection bar?
12:06:48JdGordonbecause you have voice enabled
12:06:56JdGordonyou cant hide the bar and talk the selection
12:07:12JdGordonwhen the bar is hidden selection doesnt act like it does with it visile
12:07:33pondlifeThe bar shouldn't go into the gaps...
12:07:37n1sor maybe that "feature" got lost in the list conversion :-)
12:07:40 Quit ddalton (Read error: 104 (Connection reset by peer))
12:07:51pondlifeIt should go from version straight to buffer for example, no?
12:07:52Lloreann1s: Unfortunately, it's necessary considering how often Windows apparently messes up remaining disk space.
12:08:06 Join ddalton [0] (
12:08:15JdGordonpondlife: ill have a play and see if that will work
12:08:17pondlifeCurrently highlighting Version reads the time!
12:08:25ddaltonsorry pondlife got disconnected.
12:08:28 Join spiorf [0] (n=spiorf@
12:08:34ddaltonyes exactly
12:08:36Lloreanpondlife: That's even worse.
12:08:49n1sLlorean: i think its nasty to do it that way though like a hidden button press in an info screen... also I don't see the point of that scan very often
12:08:51LloreanA visual cue contradicting the audible cue.
12:08:51pondlifeWhat's worse than whate?
12:08:55ddaltonso why don't you agree with sdoyon? what is there that confuses users?
12:08:56 Join mrkiko [0] (
12:08:56pondlifeAh, yes
12:08:58pixelmapondlife: is your voice file up to date?
12:09:01pondlifeIt's buggy at the momen
12:09:22pondlifeMaybe not
12:09:23Lloreanddalton: The whole point of the voice UI is that it Voices the UI. By very definition it's supposed to read what is on screen.
12:09:36Lloreanddalton: You're going contrary to what it's *supposed* to do.
12:09:42*pondlife builds a voice file
12:09:51Lloreann1s: Agreed, there should probably just be a debug option for "Scan Disk"
12:10:13Lloreann1s: Except it may be necessary to have a real function for it, since it'll be useful even in "release" builds should one ever happen
12:10:31ddaltonLlorean: 1. do you use voice? 2. If sdoyon is the maintainer why is he saying this? The info screen is different to the rest of rb
12:10:34mrkikoHi all!
12:10:45Lloreanddalton: No, the info screen is part of RB.
12:10:53n1sLlorean: we could put it in the "system" menu
12:10:58Lloreanddalton: "Maintainer" means "he knows the code" not "he gets to unilaterally decide a new purpose for it"
12:11:02ddaltonmrkiko; re your pm pondlife is here
12:11:07rashern1s: better yet, in "disk"?
12:11:22*pondlife hides
12:11:25n1srasher: right! ;)
12:11:31ddaltonwell why did JdGordon have the time voiced when it was over version?
12:11:35*GodEater_ throws a table cloth over pondlife
12:11:43Lloreanddalton: As well, yes, I use voice occasionally. As I said, it is not only for blind users, and it needs to take into account some users CAN see the screen, and DO know the order of things on it from sight.
12:11:46JdGordonddalton: because the time isnt displayed
12:11:47*pondlife impersonates the furniture
12:12:15 Quit pepie34 ("Ex-Chat")
12:12:22ddaltonI know. But why not voice something different. As long as the user finds out what they want to hear there happy.
12:12:23Lloreanddalton: Having "Time" voiced when "Version" is highlighted is a problem we're already discussion.
12:12:32mrkikoRegarding fs#8003: it doesn't occur without voice as asked by steve
12:12:33pixelmapondlife: outdated voice file?
12:12:34Lloreanddalton: Because not all users ARE happy.
12:12:42pondlifeJust updating
12:12:52Lloreanddalton: Why is it BAD to reorder the screen?
12:13:14ddaltonwell why not voice something different to displayed? what is so bad about this?
12:13:23rasherddalton: Because it's confusing.
12:13:26pondlifemrkiko: Thanks
12:13:30Lloreanddalton: I've said it several times now: The purpose of the voice UI is to VOICE the UI. As in "read what is on the screen"
12:13:34ddaltonwell then my hack might work.
12:13:37ddaltonre 8003
12:13:47pondlifemrkiko: Can you put a note on 8003 please?
12:14:13JdGordonanyone got a better order than ?
12:14:18ddaltonLlorean: So you think it is just as quick to find the free disk space with voice as by looking at the screen?
12:14:42ddaltonrasher: its not.
12:15:06Lloreanddalton: No, I don't. But I don't think "to make it faster" is a good reason to randomly decide to make the voice not match the screen on random screens.
12:15:13Lloreanddalton: How often do you need the free disk space *immediately*?
12:15:16pixelmaddalton: sure it is confusing, for people who use voice and can see
12:15:30ddaltonas an example?
12:15:34ddaltonno ?
12:15:35Lloreanddalton: You do not get to arbitrarily decide what is or is not confusing for other people.
12:16:01ddaltonLlorean: really? so voicing the time but not displaying it isn't confusing?
12:16:06pondlifeAs a list, shouldn't it just voice the selected item? So you could scroll down to disk info or whatever quickly.
12:16:08rasherddalton: Yes it is!
12:16:14*ddalton Wonders
12:16:19rasherddalton: which is why we're going to add the time display to the screen..
12:16:21Lloreanddalton: I said it was confusing SEVERAL TIMES
12:16:24ddaltonrasher: no its not :-)
12:16:35Lloreanddalton: Again, YOU do not get to decide what is confusing for other people.
12:16:39ddaltonLlorean: so you say its confusing and that's it?
12:16:46pondlifepixelma: It's still broken with a new voice file
12:16:48LloreanI say "It should NEVER voice something that is not displayed"
12:16:49ddaltonim not
12:16:56Lloreanddalton: You told rasher it's not. Twice.
12:17:07*ddalton Wonders if Llorean has seen the code...
12:17:08pondlifeddalton: It might be confusing because currently it's rather broken
12:17:13rasherddalton: If you can see what's on the screen, having something else voiced is confusing. And it is. End of story.
12:17:17Lloreanddalton: What does seeing the code have to do with ti?
12:18:06Lloreanddalton: For one important example why it's confusing: When a voice file is broken, it voices things differently from the screen. It is reasonable then to expect many users to be confused when they encounter a screen voiced differently than it displays, and expect to need new voice files.
12:18:24ddaltonsurely voicing the time and not displaying it is confusing then? I hear "current time" and nothing is displayed about "current time" that's really confusing is it?
12:18:25pixelmapondlife: yeah, my second thought was it's just voicing numbers, so maybe something else is wrong
12:18:41Lloreanddalton: Stop repeating that. I've told you several times now, that IS confusing. And we've all agreed to it.
12:18:52*ddalton Hasn't
12:19:05ddaltonLlorean: do you use voice
12:19:12Lloreanddalton: I've told you yes to that too.
12:19:18 Join pepie34 [0] (
12:19:44*GodEater_ uses it too
12:19:57ddaltonsorry didn't see that. Well usually when a voice file is broken it won't voice one string above or below instead it might say system for say alarm im not sure
12:20:09Lloreanddalton: Or... read something that's not displayed on screen?
12:20:09mrkikoWhile fs#8003 is active, I think it's better (at least for the H 300 port) to stabilize. Somone has hints for me on how to debug? I can't get the simulator's audio working, right now.
12:20:10JdGordonpondlife: ok, it should skip the gaps now :)
12:20:27pondlifeJdGordon: Selecting version voices the time, and the buffer size isn't voiced. The disk stats are swapped too...
12:20:46 Join linuxstb [0] (n=chatzill@
12:20:47 Join Frazz [0] (
12:20:50JdGordondisk stats swapped?
12:21:16Lloreanddalton: Simply put, though, you don't get to decide what the job of the voice UI is. And right now it's not doing its job. I really don't see why you're dead set against reordering the screen.
12:21:30Lloreanddalton: I mean honestly, reordered it could read things in the order you like, and still have voice match the display.
12:21:43ddaltonFine reorder it. But what is the blank option?
12:21:59LloreanIt's blank...
12:22:15*ddalton Thinks everyone likes to argue here
12:22:33linuxstbddalton: No we don't... ;)
12:22:35ddaltonLlorean: what's the point of that? Just to add some confusion?
12:22:43JdGordonlinuxstb: yes we do!
12:22:52*mrkiko likes confusion :)
12:22:53JdGordonddalton: it makes it easier to read
12:23:00*ddalton Agrees with JdGordon for a change :-)
12:23:27LloreanThey do make it easier to read. I don't see why you're asking, though.
12:23:58ddaltonhey being blind and coming a cross a blank entry like that isn't good. Very very confusing for those blind users that can't code.
12:24:02ddaltonor understand it
12:24:39LloreanThen come up with a solution.
12:24:47ddaltonLlorean: well can we get rid of it?
12:24:48mrkikoMay be we can add "Blank" to the voice file
12:24:58GodEater_or just get the selection bar to jump it?
12:24:59ddaltonso we can stop some confusion for the blind users!
12:25:01Lloreanddalton: We could get rid of it. We could add a descriptive voice.
12:25:07ddaltonor the voice users!
12:25:14LloreanOr we could just have the selection bar skip blank lines.
12:25:25ddaltonwhat do you mean the voice says "blank"?
12:25:38rasherLlorean: that sounds like the right solution
12:25:48Lloreanrasher: GodEater_ suggested to it, but it's the most ideal
12:26:20mrkikoddalton: when you hit a blank entry the voice may say for example "blank entry"
12:26:40ddaltonmrkiko: pointless for a voice user.
12:26:44ddaltonso just skip it.
12:27:20 Join advcomp2019_ [0] (n=advcomp2@
12:27:30mrkikoddalton: ok, we're blind and should try to adapt the project to our needs. But I don't like being invasive: if they wanted to have this blank entry, then ok, they should have it :)
12:28:06ddaltonmrkiko: yes of course. Then I might as well just skip it on a press of down or up
12:28:21mrkikoddalton: infact
12:28:27pixelmaI don't see why we would need selectable blank entries at all
12:28:34mrkikoddalton: it may confuse us if the voice say nothing but if the voice notifies that
12:28:38mrkiko... no problems
12:28:41*karashata thinks skipping blank entries would be a smart idea
12:29:01Lloreanpixelma: I think the reason we have them is more "there's not a good way to skip them in lists yet"
12:29:02mrkikoskipping blank entries is another good idea
12:29:27Lloreanpixelma: I seem to remember talk of skipping the spacers in the text viewer menu, for example.
12:29:30mrkikoHow many blind users of rockbox here?
12:29:38mrkiko(it's a curiosity, not regarding this argument)
12:29:40 Join Isolinear [0] (
12:29:43ddaltonmrkiko: im one of them
12:29:55mrkikoSo we are 2 ...
12:30:11 Quit CaptainSquid (Remote closed the connection)
12:30:16karashataI gotta ask, though... why is the info screen even a selectable list in the first place?
12:30:20ddaltonmrkiko: did you get my pm?
12:30:22pixelmaLlorean: yes, I remember that one too, and I don't like them there too.
12:30:36ddaltonto find info quicker with the voice.
12:30:44ddaltonNot successful yet...
12:30:51karashataahh, okay
12:31:00karashatathat makes sense then
12:31:12mrkikoddalton: my brain knows PM == power management only.
12:31:39*ddalton Forgets about the info screen. and works on something else
12:32:09ddaltonmrkiko: private message on irc
12:32:26ddaltondid you get it?
12:33:35ddaltondon't worry...
12:35:26pondlifeErm, did no-one notice "(11:19:55 AM) JdGordon: pondlife: ok, it should skip the gaps now :)" ?
12:36:08GodEater_pondlife: I noticed it but had no idea what JdGordon was referring to
12:36:17pondlifeThe blank entries I think
12:36:20mrkikoddalton: yes, i got the PM...
12:36:27pondlifeSorry was away for a bit
12:36:27JdGordonyes, blank entries
12:36:27mrkikoddalton: but I've not loaded the noisyquery irssi script
12:36:31mrkikoand so it didn't warn me.
12:36:51ddaltonmrkiko: what's the answer
12:37:00*ddalton Wonders if he will be kicked
12:37:19mrkikoddalton.... wait
12:37:42pondlifeJdGordon: Disk capaicity is voiced as free disk space. Free disk space isn't voiced at all.
12:39:07 Join The-Compiler [0] (
12:40:55The-CompilerI need a picture showing a Sansa c200 running Rockbox or showing the Rockbox-logo. Something like in big (yes, the filename ends with small, but there isn't any c200.png or c200-big.png)
12:41:39LloreanThe-Compiler: I'd imagine there's an .svg in the source for the manual?
12:42:35n1sThe-Compiler: if the button texts donesn't matter you can use this straight from the manual
12:42:59The-CompilerI need one without the button texts
12:43:16LloreanIf you take the svg from the manual, it's very easy to remove the button texts.
12:43:29pixelmait's a drawn one not a photo. You can open the svg and delete the text layer
12:43:35n1sman that c200 has a lot of buttons
12:43:49 Quit advcomp2019 (Read error: 110 (Connection timed out))
12:44:17The-Compilerok, i'll try it, thanks
12:44:22pixelman1s: yeah, still not easy to find a proper keymap for the virtual keyboard for example... :\
12:45:49ddaltonmrkiko: BTW did you post a comment to fly spray?
12:45:57n1spixelma: what do you use as accept action in that screen?
12:46:20pixelmacurrently it's volume up or down
12:46:53LloreanIt seems like the vkeyboard would only need six buttons.
12:47:00pixelmaI should really put my two new version up in the tracker for discussion
12:47:01LloreanUp, down, left, right, backspace, and select.
12:47:28JdGordoncan I put the date as dayname day monthname year instead of dayname year monthname date?
12:47:54pondlifeWhat does the time/date screen display?
12:48:04*JdGordon prefers "Wed 24 Oct 2007" over "Wed 2007 Oct 24"
12:48:12JdGordonthe time/date screen does the latter
12:48:19pixelmaLlorean: not when you don't use line edit mode but until now I thought it's a waste to not make use of that much buttons. With line edit mode you wouldn't even need a separate backspace
12:48:55*pondlife hates to say it but fears we need a date format option .. :/
12:49:08JdGordonnot gonna happen!
12:49:11Lloreanpixelma: How do you get away with dropping backspace? Create a virtual keyboard button for it?
12:49:21JdGordonalhough.. maybe it can be added now its actually displayed somehwerre
12:49:25*karashata thinks it would make more sense to order it dayname, monthname, day, year
12:49:48*linuxstb guesses karashata is from the US
12:49:56karashatacanada, actually
12:50:03pondlifeD/M/Y is more common, no?
12:50:12pixelmaLlorean: line edit mode... for backspace and move around the edit line you can "step" on it. Ipods should have it
12:50:12pondlifeOr is that just Europe?
12:50:38karashatad/m/y is more common for numbered dates, yes
12:51:06pondlifeM/D/Y seems silly to me - where's the sense in .. It's like writing the number 789 as 897
12:51:08linuxstbAll I know is the UK is d/m/y, the US is m/d/y. And now I know Canada is also m/d/y and .au is d/m/y...
12:51:12Lloreanpixelma: How do you confirm then?
12:51:23karashatabut for partly abbreviated names it makes more sense m/d/y
12:51:29Lloreanpixelma: My thinking would be "Select" selects letters, or when you move down to the line, confirms the filename. Backspace then could be universal.
12:52:22 Quit pepie34 ("Ex-Chat")
12:52:35LloreanThough there's not a button for backspace on iPods like that, at least. =/
12:53:11pixelmaLlorean: you usually have a different button for confirm or abort. Select is backspace when you are on that line. On Ondio for example "mode" is backspace and "long mode" is accept
12:53:31 Join CaptainSquid [0] (
12:53:40n1spixelma: i would either use long press of select as accept or make a button on the keyboard for it
12:54:10Lloreann1s: I'd rather "move to the filename, and then press select" as accept, in most cases, as it makes it harder to accidentally insert one extra character.
12:54:13pixelmamore precise "mode" is backspace on the line or select when over the letters, "long mode" is accept everywhere
12:55:04pixelmaI like how it is now
12:55:08amiconnLlorean: I'd rather want accept to be universal (as it currently is) than backspace
12:55:10n1sLlorean: if the normal insert action is on release it should not be a problem, and I think pressing select in line edit mode makes no sense for accepting
12:55:23Lloreann1s: It's not a problem for people whose fingers are steady.
12:55:26amiconnYou always need accept to enter something, but you do not always need backspace
12:55:55kugelcouldn't there be a shortcut, like rec+select, to accept?
12:56:20LloreanI have no problem with the way things are now, just saying I personally would have a slightly different preference.
12:56:23amiconnButton combos should only be used if there is no way to get away with short/long presses
12:56:32LloreanButton combos should be avoided, yes.
12:56:39*Llorean can't believe there's now also a double click patch.
12:56:46amiconnShort/long allows one handed operation, button combos do not
12:57:21pixelmakugel: I also don't like button combos and they are also not possible everywhere (on the different targets) and in every combination
12:57:32amiconnWay back before we had short/long presses, some actions on the Player even had 3-button combos :>
12:57:45kugelok, I just had this idea
12:58:20kugelbut long press is indeed more usable
12:59:01LloreanAnyone want to reject the double click patch? It seems to me it's not a wanted feature, but a) I could be wrong on this one, and b) I have a feeling the author will pester me about it if I close it.
12:59:37JdGordonok, who's game enough to test the patch out?
12:59:38GodEater_is that a "please will someone other than me close it?" or a check to see if people agree that it should be closed ? :)
12:59:56GodEater_well I don't want it either
13:00:05JdGordonLlorean: I actually like the idea
13:00:09Lloreanyegods is one of those "people who think the fact I found problems with their idea means I have something against them personally"
13:00:29GodEater_what does he use the patch for ?
13:00:35LloreanSeek-mode on iPods.
13:00:42GodEater_seekmode ?
13:00:46GodEater_wth is that ?
13:00:59LloreanDouble click select, then you can seek to a location in a file by rotating on the wheel
13:01:06JdGordonpondlife: ddalton: anyone else:">
13:01:22LloreanIt apparently interferes with pretty much any other normal function of the device, assuming you press keys quickly
13:01:28*GodEater_ wonders why that's more useful than holding the ff/rw buttons
13:01:32 Quit AlexC (Network is unreachable)
13:01:35 Quit mrkiko ("leaving")
13:01:38LloreanGodEater_: People are used to it on iPods.
13:01:45GodEater_I'm not :)
13:01:50LloreanGodEater_: I wonder why it's more useful than a "seek mode" option in the WPS context menu
13:01:58GodEater_yeah that too
13:02:13GodEater_so does his patch include both features ?
13:02:14LloreanThe feature doesn't need double click, and it saves maybe at best half a second to double click select rather than "select, up, select"
13:02:23JdGordonI tihnk the feeling is you can get more accurate seeking with the wheel? or faster or something?
13:02:32LloreanGodEater_: Yes and no. There's one patch for double-click on iPod Video, then a second for Nano + seeking mode
13:02:41LloreanJdGordon: What use do you see for double click?
13:02:47n1smy problem with the double click is that someone wanted it on the left and right buttons in wps to do something in which case they would have short, long, short-long, and double which is waay too much IMHO
13:02:47GodEater_so if we want seeking, we'll have to prune it out
13:03:03kugelimo turning a wheel feels better than holding a button
13:03:05LloreanGodEater_: Or close the task and demand he resubmit according to the guidelines.
13:03:09JdGordonLlorean: that use, and the fact it triples the button count on ever target
13:03:24kugeland it might be indeed faster, when you add the scroll wheel acceleration
13:03:58GodEater_talking of which, I vote once again to get 7738 commited ;)
13:04:00LloreanJdGordon: Not really. There are many places you'll want to be able to press buttons fast succession anyway. Skipping multiple times, browsing, etc.
13:04:00ddaltonwhat's the link again? JdGordon?
13:04:09n1sIMO its unnecessary ui complication">
13:04:42LloreanJdGordon: As far as I can see, any feature you put on it will either A) annoy people who can't press the button fast enough, or B) Annoy someone who wants to be able to make single presses faster.
13:05:00GodEater_I think those are both good reasons to reject it
13:05:06JdGordonyeah, probably
13:05:17LloreanI do agree it'd add a few more buttons.
13:05:30LloreanBut, I think short-long already does that job in the one place it's useful so far, right?
13:05:51JdGordonwe could add it and make it configurable...
13:05:54JdGordonor just reject it
13:06:15LloreanI'd say reject it, or even "invalid" it, and point out "Rockbox already has a short-long double click equivalent"
13:08:02ddaltonJdGordon: 1. Have the time and date as the same option. So in the same line
13:08:05ddaltonmore to come...
13:08:24JdGordonit does that
13:08:33*pixelma is really scared and discouraged by looking at the button action definitions in metronome.c :(
13:08:53pondlifeShort-long isn't ideal though
13:08:54JdGordoni get it.. i get it.... pla sucks
13:09:21Lloreanpondlife: It's better than double click, though.
13:09:41pondlifeI was wondering... maybe it depends on the implementation.
13:09:50ddaltonJdGordon: oh ok sorry... so what order? it think it should be bat level time/date free disk space then what ever else version then buffer size...
13:09:50LloreanDouble click would interfere with multiple skips, while short long would only interfere with skip then seek, for example
13:09:58ddaltonwhy not voice the time for the blank option?
13:10:04pondlifeCurrently short/long results in the short action being processed first, right?
13:10:16GodEater_ddalton: because that would be confusing
13:10:19pondlifei.e. you actually get short action, then short/long action
13:10:21pixelmaJdGordon: metronome.c is the worst I've seen so far. Can't really make heads or tails of it, maybe I need to stare longer...
13:10:21Lloreanpondlife: But it's only used in a situation where that doesn't matter.
13:10:34pondlifeMildly annoying though :)
13:10:40LloreanFair enough.
13:10:54LloreanBut I'd really rather *nothing* depended on timing, anyway
13:10:54JdGordonddalton: the order isnt being discuessed.. is everything voiced ok?>
13:10:58LloreanI don't even like short-long, honestly
13:10:59 Join mrkiko [0] (
13:11:06ddaltonJdGordon: I see a comment that I wrote in there :-)
13:11:09LloreanThese are consumer electronics, not videogames
13:11:27ddaltonhang on...
13:12:12ddaltonI don't think anything will be heard when line 199 is executed
13:12:16pixelmaJdGordon: to be fair, I think TiMiD's last additions made it worse
13:12:38JdGordonwhat was that?
13:12:41ddaltonmight as well speak the buffer as well.
13:12:47pondlifeOf course
13:12:54ddaltonso no confusion
13:13:01JdGordonbuffer is speaking, i took that from sdoyans patch
13:13:13JdGordonnothing is spealing on my sansa though... maybe a really outdated voice file?
13:13:13pondlifeJdGordon: Sorry, still compiling here
13:13:29JdGordonthe [atch i upped doesnt have the buffer talking though
13:13:29ddaltonshould actually say hour and min now I think about it. people won't be confused
13:13:36JdGordoncan someone email me english.voice for the sansa?
13:14:09*JdGordon forgot we had daily voice updates
13:14:36ddaltonlooks pretty good to me. did you apply some of my patch?
13:15:48ddaltonJdGordon: I don't want to resync all my stuff but looks like everything will talk accept buffer of course and maybe the date case
13:15:50***Saving seen data "./dancer.seen"
13:16:00pondlifeBuffer should talk
13:16:16ddaltonmaybe I missed something
13:16:22pondlifeI mean, it ought to
13:16:32pondlifeI've not tested any code yet ;)
13:17:29ddaltonit should but won't with the current patch.
13:18:41 Quit qweru ("moo")
13:19:09JdGordonits not talking for some reason :')
13:19:11JdGordon:'( even
13:19:22ddaltonwhat isn't?
13:19:31JdGordonthat bloody screen
13:19:41ddaltonthe whole thing?
13:19:42pondlifeI'm just building a fresh voice file
13:20:59pixelmaJdGordon: he changed the plugin to multiscreen and the controls are now differently (new to me was that you need to start "playing" the beat instead of just let go of the tapping. And the code added in r14051 looks messy and it's easy to get conflicting definitions
13:21:35JdGordonah ok
13:24:46pondlifeJdGordon: The bar is now skipping the blank lines nicely, but no voice at all... same as you're seeing, right?
13:24:57mrkikoWhat plugins does rockbox load by default?
13:25:01JdGordonis it running off the end of the list?
13:25:18pondlifeIt wraps as I'd expect
13:25:25JdGordonwhich target?
13:25:28pondlifei.e Free disk space -> version
13:25:31pondlifeH300 sim
13:25:39JdGordonah! no hotswap
13:25:50*JdGordon thought the bug was elsewhere
13:26:38ddaltonmrkiko: lots why?
13:26:47Lloreanddalton: Not quite true.
13:26:48pondlifeJdGordon: Where's the info screen code?
13:27:03Lloreanmrkiko: Rockbox doesn't load *any* plugins by default, but when you install it several are on disk.
13:27:04ddaltonLlorean: what is true then?
13:27:11*JdGordon wants to move it
13:27:14Lloreanddalton: No plugins are loaded into memory until you run them.
13:27:29LloreanAt which point only one plugin is ever loaded at once.
13:27:29ddaltonwell that come with rockbox I thought was the question sorry
13:27:33mrkikoI understood
13:27:52ddaltonarguements start quickly here :-)
13:27:56mrkikoLlorean: thank you
13:27:57ddaltonnearly another one
13:28:30*karashata thinks some people are way too picky about other people being slightly less technical about how they refer to things...
13:29:00Lloreankarashata: And if someone had thought Rockbox loads all the plugins at once, we'd start getting feature requests for "not loading all the plugins to have more buffer for audio" again.
13:29:12pondlifeIt has happened :)
13:29:14LloreanIt's rather important that people are told the truth...
13:29:16 Quit maddler (Read error: 113 (No route to host))
13:29:39mrkikohey guys: It was my fault
13:29:45karashataahh, that is true, and you'd expect most people in this channel would know the difference between a plugin being installed vs. a plugin being loaded
13:29:46mrkikoI misused the "load" term
13:30:19mrkikobut I referred to "load in memory"
13:30:23mrkikonot "on disk"
13:30:25Lloreanmrkiko: Which is the proper use.
13:30:29Llorean"in memory" that is.
13:30:49JdGordonpondlife: i dunno why its not voicing... ddalton was that screen ever voicing correctly?
13:30:57pondlifeJdGordon: For some reason it keeps going back into info_speak_item
13:31:00ddaltonwhat one?
13:31:05pondlifeIt never gives itself time to talk
13:31:13JdGordonthe info screen.... the one we have been tlakign abot for the last hour
13:31:35pondlifeProbably related to the time update
13:31:40ddaltonJdGordon: I stopped listening after a while. Got board. Yes it di. Ill try it later if I get time
13:31:46ddaltonyes it did
13:32:26pondlifeJdGordon: Where's the code that refreshes the time?
13:32:43ddaltonis there a call to get_yime?
13:32:57JdGordonget_time sholdnt mangle the voice though
13:33:27ddaltonJdGordon: I will have to test tomorrow.
13:33:34mrkikoJdGordon: are you trying to resolve 8003?
13:33:43pondlifemrkiko: No
13:33:44ddaltonim up at 6 30 so that is when I can tell you
13:33:47*mrkiko forgot to type fs#
13:35:26JdGordonbeer n pool time
13:35:29ddaltonJdGordon: what's the link and I will test
13:35:49JdGordonddalton: im sure your client can backtrace the convo
13:36:06ddaltonit can but takes for ever...
13:36:13pixelmaJdGordon: metronome.c defines a combination of 2 for some targets even 3 contexts now... IIUC
13:36:26JdGordonpixelma: feel like rewriting it :)
13:36:44JdGordonpondlife: ddalton:"> im heading out, but if you can figure out why its not talking it would be good
13:36:47*JdGordon gone
13:39:59ddaltoncan't get it test later
13:40:55mrkikoddalton: what is the name of the braille display you're using?
13:49:15 Join Gio [0] (
13:51:36mrkikohi gio :)
13:51:45Giohi mrkiko
13:51:50 Quit ddalton ("leaving")
13:52:21 Join joshin [0] (n=joshin@VDSL-130-13-11-27.PHNX.QWEST.NET)
13:52:21 Quit joshin_ (Connection reset by peer)
13:53:59 Quit iamben (Read error: 104 (Connection reset by peer))
13:54:00 Quit Soap ("ChatZilla [Firefox]")
13:56:15 Quit karashata ("Leaving.")
14:00:28 Join karashata [0] (
14:08:17 Part mrkiko ("RETURNING IN MY OWN WORLD!")
14:11:10amiconnlinuxstb: around?
14:12:47amiconnDo you know whether it's important that the ape filters use int16_t?
14:13:10amiconnIf we could go int32_t for arm, this would allow further savings for arm7
14:13:45amiconnIt would be neutral for arm9 and higher - and arm9 could use the arm7 asm then because that'd also be neutral
14:14:22amiconn(the int16_t asm optimised for arm7 would *slow down* arm9 and higher)
14:17:24linuxstbI don't know, we would have to test it.
14:22:51pondlifeJdGordon: Hmm, you closed #8012?
14:23:07pondlifeI'll reopen with a more talkative patch
14:24:12Gio bye
14:24:14 Part Gio ("Konversation terminated!")
14:26:39 Join JRoT|koken [0] (
14:26:53 Nick JRoT|koken is now known as JRoT (
14:27:13pondlifeJdGordon: OK, it's up at
14:27:17 Nick JRoT is now known as JRoTje (
14:28:01 Quit JRoTje (Client Quit)
14:28:33 Join JRoT [0] (
14:29:58 Join J3TC- [0] (
14:32:58pondlifeJdGordon: Oops, sorry - committed, but didn't give you credit.
14:33:18pondlifeLet me know if there should be any changes to CREDITS too.
14:33:41pondlifeBack in a bit.
14:36:43 Quit donutman25 (Read error: 104 (Connection reset by peer))
14:37:31 Quit XavierGr (Nick collision from services.)
14:37:34 Join XavierGr [0] (
14:38:35LinusNpondlife: my mixed-case identifier alarm triggered: InfoScreenOrder
14:38:42 Quit Frazz (Read error: 104 (Connection reset by peer))
14:44:00 Join Frazz [0] (
14:47:03 Join webguest43 [0] (i=5848e008@gateway/web/cgi-irc/
14:48:49*JdGordon back
14:49:32JdGordonLinusN: thats my code.. its not actually used anywhere so didnt put much effort into the name
14:51:34JdGordonpondlife: im happy to not get credit... you get the rap fr the red delta instead of me :D
14:52:31 Quit webguest43 ("CGI:IRC (Ping timeout)")
14:53:03 Join mf0102 [0] (n=michi@
14:53:06pondlifeJdGordon: Doh!
14:53:16pondlifeCan you shrink it at all?
14:53:31JdGordonnot really
14:54:03pondlifeMuch bigger delta than I'd have expected :/
14:54:07linuxstbJdGordon: What isn't being used? The enum seems to be used, with mixed-case values...
14:55:13JdGordonwait.. whats wrong with mixed-cased values? (what are mixed-cased values?)
14:55:34pondlifeWe avoid CamelCase
14:56:50GodEater_and hungarian notatino
14:56:51JdGordonhmm.. /me forgot about that.. I like CC for enum values
14:56:55GodEater_notation too
14:57:07pondlife"Variables and function names should be all lower case." from CONTRIBUTING
14:57:20JdGordonhahah sif read CONTRIBUTING :d
14:57:35JdGordonI dont tinhk anyone bakcs hungarian anymore do they?
14:57:44GodEater_except M$ :)
14:58:02JdGordonno, even they dropped it, I read a c# book and they said MS dropped it also
14:58:07pondlifeJdGordon: If you fix it, you get to look like the good guy to my bad guy.
14:58:30*JdGordon needs to sober up a bit before doing any commiting
14:59:36pondlifeAh, your SVN has a breath pipe attached too?
15:00:36JdGordoncondition of access im afraid :'(
15:01:39pondlifeWhy did you move that stuff out of screens.c, btw?
15:01:51JdGordonwhich stuff?
15:02:03JdGordonthe 2 arrays?
15:02:30JdGordonbecause they were only for bitmap lcd in screens.c, whereas in main_menu.c they are for all rtc
15:02:52JdGordonyou forgot to fix bjorns (c) notice...
15:03:07*JdGordon wathcing some tv.. back in 45
15:03:17pondlifeI just committed your patch... :)
15:08:12 Join nicktastic [0] (n=nick@unaffiliated/nicktastic)
15:09:01 Nick parafin|away is now known as parafin (
15:09:24 Join zicho [0] (
15:09:39 Join webguest77 [0] (i=5848e008@gateway/web/cgi-irc/
15:15:07 Quit webguest77 ("CGI:IRC (Ping timeout)")
15:15:51***Saving seen data "./dancer.seen"
15:20:16preglowmy, our resampler sounds crap with 8khz files
15:21:07preglownot far from it
15:23:59 Join nicktastique [0] (n=nick@unaffiliated/nicktastic)
15:24:04 Quit J3TC- (Read error: 104 (Connection reset by peer))
15:24:44 Join J3TC- [0] (
15:24:59 Quit nicktastique (Client Quit)
15:25:50*preglow has a stupid bug in his code
15:28:44LinusNreally? wow!
15:31:10preglowyeah, this is quite a shock to me
15:31:59LinusNi'm glad that never happens to me
15:34:20GodEater_you kids....
15:37:51preglowLinusN: didn't you do some emac work on binutils?
15:38:03LinusNyes i did
15:38:15*Zagor just found out I can send 16 KB packets to the usb controller. whopee!
15:38:22preglowit looks like msac.w with parallel load is broken, would you be able to either check it out or give me some pointers?
15:38:32Zagorthat ought to spice things up a bit
15:38:44*preglow looks forward to the next transfer speed measurement...
15:38:47LinusNpreglow: all i did was add a few register names, never messed with the encoding
15:38:52preglowLinusN: ah, right
15:39:57markunZagor: what's still missing from the usb code?
15:39:58preglowwould any of you guys know if there exists any other widespread calling conventions than the usual ones for m68k and coldfire?
15:40:01Zagorpreglow: ...and then I can link a whole bunch of 16-kb transfers together (as many as the ram can hold) and the controller follows the link automatically. that ought to max it out pretty good.
15:41:03Zagormarkun: lots. I'm still in the test-code phase. my optimism the first couple of days was a bit naive :)
15:41:05preglowZagor: will that also eat 16 kb of iram for the data?
15:41:17Zagorpreglow: no, I use dram
15:41:23preglowoh, i thought you had to use iram
15:41:24Zagorin fact I plan to use the mp3 buffer
15:41:25preglowwoopee, then
15:41:52Zagorno, it just has to be uncached.
15:43:08peturaha, so only the structs need to be in iram then?
15:43:48Zagorno nothing needs to be in iram, it just needs to be uncached
15:44:32linuxstbZagor: Is CPU speed very important with regards to transfer speed? I mean will Rockbox need to boost the CPU whilst in disk mode?
15:45:48Zagorlinuxstb: I don't think so. the cpu only acts as a middleman doing small stuff like setting up transfer descriptors. the big work is done by the drive/flashcard and the intelligent usb controller.
15:46:50 Join barrywardell [0] (
15:47:15Zagorof course faster cpu might result in slightly faster disk access, which might improve total throughput. but I think it's far from necessary
15:48:43 Quit Mouser_X (Read error: 110 (Connection timed out))
15:54:20pondlifeLinusN: Hope that's better...
15:58:04 Join Gnu47 [0] (
15:58:16 Quit Zagor ("Client exiting")
16:03:44 Join Nico_P [0] (n=nicolas@rockbox/developer/NicoP)
16:04:07Nico_Ppondlife: hi
16:04:07 Join scorche|w [0] (n=8dc5049d@rockbox/administrator/scorche)
16:04:12pondlifeHi Nico_P
16:04:35pondlifeI've not found many problems :)
16:04:43Nico_PI've seen you've had rebuffering issues
16:05:11pondlifeOnly that it no longer uses the ata callback,
16:05:41pondlifeSo it doesn't notice an unrelated spinup
16:05:59Nico_Pwell hopefully it wstill behaves the same, but doesn't wait as long before filling
16:06:19Nico_Pit should notice any spinups
16:06:32pondlifeNo, it never buffers unless it's low on data
16:06:42pondlifeTry the recipe I put on the wiki
16:07:04 Join bluebrother [0] (i=GweZxw09@rockbox/staff/bluebrother)
16:07:09Nico_PI find that very suprising. I've seen it rebuffer several times on my gigeabeat
16:07:34Nico_PI'll check again
16:07:40pondlifeOn H300, maybe the ATA status code is a bit faulty?
16:07:48pondlifeAnd wasn't used much until now?
16:08:16pondlifeI assume it's intended to rebuffer whenever the disk is in use?
16:09:03Nico_Pthe commit I did to replace the callback wasn't supposed to change that behaviour, but maybe I made a mistake somewhere
16:09:27pondlifeYou need an early callback really
16:09:38Nico_Pyes, that would be the best
16:10:11Nico_Por I could chain buffering request in some wya, but I like that idea less
16:11:12Nico_Pdid you see the commit related to the conf_ vars?
16:11:33LinusNpondlife: much better :-)
16:12:28 Quit CaptainSquid ("Miranda IM!")
16:12:42 Join KiOui [0] (
16:13:06KiOuiHi there: I think I found why the Rockbox UI sometimes crash on iPods
16:14:23KiOuiIt seems that filename length is part of the bug. I tried to play a quite long playlist (more than 6 hours) with short filenames (that fits into the screen) and it works quite well.
16:15:13Nico_Pargh I can't exit the properties screen :/
16:15:29KiOuiBut if there is a long filename in it, it seems to lock. The key is still semi-handled because the LCD lights up. But no track change / volume / menu / ...
16:16:09linuxstbHow long is the long filename, and does it have any non-ASCII characters?
16:17:12 Join desowin [0] (
16:18:36KiOuilinuxstb: No idea of the length because I'm blind and unable to count characters. But yes it shas non-ASCIII chars
16:19:07Nico_Ppondlife: I know why we don't quite agree on the bug
16:21:37Nico_Pyou have several tracks buffered, right?
16:21:51pondlifeProbably, yes
16:22:09pondlifeI'm just saying "one third useful"
16:22:16 Nick fxb__ is now known as fxb (
16:22:41Nico_Pwhat's happening (probably exactly like before) is that even though only one third is useful, the buffer is still full. fill_buffer() gets called but can't do anything
16:23:02Nico_Pwhat we *could* do is call the low buffer callback to ask the audio thread for new tracks
16:23:31 Join bluebrot1er [0] (
16:26:51 Quit bluebrother (Nick collision from services.)
16:26:55 Nick bluebrot1er is now known as bluebrother (
16:29:30pondlifeNico_P: We should fill the buffer based on useful data only
16:29:31Nico_Ppondlife: if you use a track that's bigger than the buffer you'll see rebuffering
16:29:55Nico_Pyeah probably. then the callbacks have to be called on low buffer
16:30:14pondlifeWhich callbacks? The unbuffer ones?
16:30:44Nico_Pno, the "low buffer" ones, which should probably renamed to "request data"
16:30:54pondlifeAh, yes.
16:31:08Nico_Pthe difference with before is probably that playback.c had an idle ata callback to add new files and I removed that and replaced it with a low buffer callback
16:31:09pondlifeAren't they called on low buffer anyway?
16:31:09KiOuilinuxstb: Do you think non-ASCII chars may cause the bug?
16:31:19 Join Mouser_X [0] (n=someone@
16:31:46Nico_Ppondlife: they're called on low buffer only. they should also be called when the useful level is below the high watermark and the disk is spinning
16:32:03Nico_Pit's a trvial change
16:32:11pondlifeWell, whenever we want to fill
16:32:26pondlifeI'd hope it would come out in the wash..
16:32:35pondlifeDid you see the auto_change_dir bug?
16:32:41pondlifei.e. did you repro it?
16:32:59 Quit desowin ("use linux")
16:32:59Nico_Pah no, forgot about it
16:33:02linuxstbKiOui: All I know is that non-ASCII chars take up more bytes than ASCII (Rockbox uses utf-8 internally), so if there is a problem with long filenames, it will be easier to hit the limit with non-ASCII chars. I don't know what, if any, checks Rockbox does on filename lengths.
16:33:22Nico_Ppondlife: did you see the commit for the conf_* vars?
16:33:36Nico_Phow do you like it?
16:33:37pondlifeI'm still not sure how they work!
16:34:14pondlifeNot much
16:34:32 Quit bluebrother ("bbl")
16:34:41pondlifeI probably don't understand, but I thought those were codec requirements, and hence should only affect playback.c, not buffering.c..
16:34:45KiOuilinuxstb: Do you have an idea of max filenames lengths in Rockbox?
16:34:53KiOuiIs it target-dependant?
16:35:02pondlifei.e. They affect the flow of data out of the buffer, not the reading from disk..
16:35:04JdGordon260 chars iircs
16:35:15Nico_Ppondlife: they give the codec control over how the buffering works
16:35:16pondlifeBut, like I said, I don't really understand them
16:35:56pondlifeWhat if I am buffering a mixture of SIDs and MP3s, say...
16:35:57Nico_Pthe codec apparently wants to contol the watermark
16:36:15Nico_Pyeah I know, but for now I simply translated how they work in SVN (improving it a bit on the way)
16:36:44linuxstbJdGordon: Isn't it 260 bytes, not chars?
16:36:52pondlifePersonally, I'd scrap it and see if anyone notices :)
16:37:07JdGordonyeah, 260 bytes, for the whole path string.. not just the filename
16:37:33pondlifeUntil someone can explain why the codec cares about disk->buffer, rather than buffer->codec....
16:37:37JdGordonwelll... MAX_PATH which is either 260 or 290, cant remember which
16:37:48KiOuiJdGordon: I think I'm not reaching the max length
16:38:10pondlifeJdGordon: Fixed up your varnames, can you fix up Bagder's (c)... I don't think my editor deals with non-ASCII properly.
16:38:25JdGordonits zagors, and yeah
16:38:30pondlifeJdGordon: Ah, yes
16:38:46JdGordonand _my_ var names??? yo commited it :D
16:38:58pondlifeHah, they're my vars now :)
16:39:04Nico_PI'm not sure
16:39:58pondlifeIt seems wrong to me that playing a SID would affect the buffering of a future playing MP3...
16:40:02pondlifeOr whatever
16:40:08JdGordonwould anyone have objections to splitting up the different screen codes into a new folder under apps/ (screens maybe?) screens.c is nasty (as is debug_menu.c) and the ino screen shodlnt really be in menus/main_menu.c
16:40:25Nico_Ppondlife: I see what you mean, but I'd prefer to take care of that after having committed the main chunk ;)
16:40:34pondlifebuffering.c is just like a file preloader.. playback.c should deal with the codec requirements...
16:40:54pondlifeHave a look at that auto-dir change case then, I dare you :p
16:41:08Nico_PI will
16:41:32Nico_PI've solved the USB bug too...
16:41:57*Mouser_X listens to a mix of NSFs, SPCs, and MP3s (and other large "streamed" codecs) frequently.
16:42:32Mouser_XHow would that effect Rockbox with your new code/buffer stuff?
16:43:18GodEater_Nico_P: have you been anywhere near the crossfading code yet ?
16:43:33Nico_PGodEater_: very briefly
16:43:47GodEater_still broken then ? :)
16:43:53Nico_Pbut playing tracks with crossfade enabled has been no problem for me
16:44:03Nico_Ponly enabling crossfade during playback fails badly
16:44:10GodEater_but enabling or disabling it breaks for me
16:44:25JdGordonKiOui: you may have stumbled on the bug where for some unknown reason the list code crashes (?) with long-ish lines.. i tihnk we found the max it could acually display was 160 or so
16:44:34JdGordonbut that was ages ago.. I thought it was fixed
16:45:43KiOuiJdGordon: I think the bug is the same as reported onto the ipodstatus wiki page (issue with 5.5g models)
16:45:55KiOuiThe "occasionnal crash"
16:46:16pondlifeIs this the voiced list problem?
16:46:24pondlifei.e. #8003
16:46:40KiOuiAnd it seems to depend on what we're playing: because a specific song makes the bug occur, while another one doesn't
16:47:14JdGordonmaybe its a dodgy file then and the codec is dieing?
16:47:30KiOuipondlife: I'm actually using Speech feedback; but don't know if it comes from this or not
16:47:42pondlifeIT does seem to be only when speech is enabled
16:48:02pondlifeOr so has been reported
16:48:07JdGordonpondlife: did you get voicing working in the info sceen before commiting?
16:48:19KiOuiJdGordon: I don't think so, because the file can be played normally. It is like the iPod was in Lockedmode.
16:51:20pondlifeJdGordon: Yes
16:51:33pondlifeWe lost the real-time updating clock etc though
16:51:53pondlifeThe problem was that you were doing lots of redrawing, so the voice never got started
16:52:15JdGordonah pl
16:52:27pondlifeThe code to skip the gaps was previouosly returning ACTION_REDRAW too often
16:53:45JdGordonbarrywardell: you round?
16:54:14JdGordonFUCK :'(
16:54:21*JdGordon is too tired
16:55:15 Join desowin [0] (
16:55:31markunJdGordon: go to sleep
16:55:46JdGordongotta recover my sansa first :'(
16:55:54barrywardellJdGordon: yep!
16:55:58*JdGordon accidenrtly wrote the mi4 to the bootloader instead of the .bin
16:56:19JdGordonbarrywardell: do i need to do the full recover with e200tool to get it going again?
16:56:59barrywardellyou used sansapatcher -bl PP5020.mi4?
16:57:07JdGordonyeah :(
16:57:15*JdGordon makes a point to not that its 1am here
16:57:18 Join Rondom [0] (
16:57:35barrywardellyou need to use "e200tool recover BL_SDboardsupported.rom"
16:57:39barrywardellor whatever the file is called
16:57:44JdGordonoh goody
16:57:49JdGordonok, off to find that file
16:57:58barrywardellthen copy over the BL_SDboardsupported.rom to the 16MB disk that appears
16:58:02barrywardellit's on Bagder's site
16:59:09 Part LinusN
17:01:04amiconnpondlife: Imo the info screen gaps should not exist on low line count screens
17:01:08*JdGordon hates e200tool :'(
17:01:29pondlifeIMHO there shouldn't be gaps at all
17:01:32JdGordonpondlife: hehehe, im goign to let you do commits for me more often :)
17:01:33amiconnI didn't try this yet, but on Player they will be plain annoying
17:01:51JdGordonna, the gaps are good.. but yes, on player they are bad
17:01:52pondlifeWere they not already there?
17:02:07JdGordonyes, they were always there
17:02:24amiconnI can imagine that the gaps are annoying on other small screens as well, e.g. archos bitmap or the iriver remotes
17:03:06amiconnJdGordon: I don't think they were always there
17:03:09pondlifeJdGordon: Go on, scrap them
17:03:24barrywardellJdGordon: but without e200tool, your sansa would be a brick right now.
17:03:44JdGordontrue.. but the bloody thing is tining out during the read :'(
17:04:30barrywardellon linux?
17:04:34JdGordonamiconn: pondlife: i wont really be against removing them, but they do make the screen easier to read
17:04:36 Part KiOui
17:05:19amiconnJdGordon: Only if all lines fit without scrolling
17:05:26barrywardellit sometimes takes a couple of tries before it works...maybe we can get Zagor to look at that once he's done adding USB to rockbox :P
17:06:21JdGordonor maybe i just modify the e200rinstaller to fix the bootloader instead? :)
17:06:32JdGordonits a simple write isnt it?
17:06:42barrywardellyeah, just write to the start of the hidden partition
17:06:51barrywardelland include the PPBL header
17:07:18barrywardellwhich is 0x200 bytes
17:08:11pondlifeLooks ok without the gaps
17:10:31preglowargh, i give up
17:12:33JdGordonbarrywardell: the ppbl header is 0x200 bytes? or the ppbl section?
17:14:06barrywardellthe header. see step 2 and 3 here:
17:14:40JdGordonyeah, im looking there
17:14:47*JdGordon is definatly too tired for this :p
17:15:38barrywardellthe header is 1 block (0x200 bytes), starting with "PPBL", then little endian length of the bootloader binary, then little endian magic number 0x00010000. the rest of the header is 0's
17:15:54***Saving seen data "./dancer.seen"
17:20:00JdGordonbootloader.bin is 100k, too big to send in manufac mode?
17:20:23 Join ackbahr [0] (
17:21:04pondlifeHmm, we already assume that the LANG_ constants for months and week day names are continguous and in sequence, so do we really need to look it up...?
17:21:44JdGordonwe dont assume that... but it sounds like a valid assumption anyway
17:21:54pondlifeWe do.
17:22:05pondlifeNot in the info screen, but elsewhere
17:22:33pondlifeFound it?
17:22:35barrywardellBL_SD... is 387K and I sent that fine, so...
17:22:56JdGordonkeep getting -71 protocol error
17:22:56pondlifeI can't recall, and I've just spread that assumption in an attempt to reduce bin size
17:23:56pondlifeHmm, 200 bytes...
17:24:04pondlifeProbably worth it
17:25:23 Quit ackbahr (Remote closed the connection)
17:27:08barrywardellJdGordon: I wonder why e200tool works for me but not you? Have you tried a different computer?
17:27:45JdGordoni had trouble last time i needed to recover it also, im goign to give up for now though, ill try my laopt tomoorw
17:28:45barrywardellmeanwhile, maybe sansapatcher should be changed to only accep an mi4...
17:29:02JdGordonno... it should be changed to NOT accept mi4 for -bl
17:29:12barrywardellor that
17:29:47barrywardellbut if it only accepted the mi4, then we would have the header to validate against
17:32:55 Quit JdGordon ("Konversation terminated!")
17:33:48amiconnpondlife: Simply use that assumption. It is meant to hold. Same goes for several other things, e.g. the voice clips for latin letter, the clips for digits etc
17:33:58 Quit J3TC- (Read error: 110 (Connection timed out))
17:34:44pondlifeOK, it's committed
17:35:17pondlifeHope I didn't break the copyright notice again..
17:36:56*Nico_P whispers unicode
17:37:16*pondlife has a nasty habit of using notepad :(
17:38:32Nico_Pthough I'm saying unicde because I think the source files should be unicode
17:38:36*scorche|w whispers to pondlife how to make notepad2 the default
17:39:09 Quit XavierGr (Nick collision from services.)
17:39:10 Join XavierGr_ [0] (
17:39:11pondlifeI prefer Notetab Lite actually... must put it in my path
17:39:40Nico_Pthere's a path in windows?
17:40:09*scorche|w threatens CGI:IRC
17:43:43pondlifeamiconn: Got rid of the gaps, and saved 200 bytes. hope you like.
17:44:21pondlifeOK, back to MoB
17:50:24 Quit linuxstb (Read error: 110 (Connection timed out))
17:50:29 Quit atsea-32 (Read error: 104 (Connection reset by peer))
17:51:42*Nico_P too, back to mob
17:51:48Nico_PI'm having plane problems
17:51:59Nico_Ppondlife: it rebuffers now
17:53:33 Join alteregoa [0] (
17:55:24Nico_Ppondlife: is it worth changing the callback names?
17:55:35 Join Alexandro5 [0] (i=4fb36cb2@gateway/web/cgi-irc/
17:57:15Alexandro5when i enter e200 with rock to Files section it shows all my folders like : .rockbox ,rocks etc. is it bug or normal thing
17:57:35barrywardellthat's normal
17:58:18Alexandro5thanks, one more thing how i know when my battery is full while my device is still connected to computer?
17:58:53alteregoai fnord a question
17:59:05Alexandro5btw i'm not at rockbox mode ATM (im chargin battery in normal mode)
17:59:05alteregoaif i use rockbox with 160gb ipod, is it compatible
17:59:19alteregoai don't know if the new 160gb ipod classic is capable of rockbox
17:59:36karashataalteregoa: it's not
17:59:57 Quit spiorf (Remote closed the connection)
18:00:21scorche|walteregoa: you can see all the supported units on the front page
18:00:42karashataand likely won't be, Apple has changed their hardware a bit and seems to have encrypted the firmware...
18:00:47Alexandro5when im chargin my battery it shows battery is full but when im disconnect it from my computer the battery is almost done... someone help me please?
18:00:53Alexandro5E200 Device
18:00:54alteregoayeah i just saw 5.5
18:01:15karashatathe iPod classic is the 6th gen
18:01:34scorche|walteregoa: did you miss the "not..." part?
18:01:37 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
18:01:50karashatathe 5.5 gen is, if I'm not mistaken, the 80 GB iPod video
18:01:56barrywardellAlexandro5: if it says the battery is full, then it has finished charging
18:01:59scorche|wand 30
18:02:10amiconnkarashata: Almost. There are also 30GB 5.5th gens
18:02:19*karashata nods slightly
18:02:21Alexandro5but when i disconnect it the battery is on 10%
18:02:37barrywardellin rockbox?
18:02:54 Quit Alexandro5 ("CGI:IRC")
18:03:01 Join Alexandro5 [0] (i=4fb36cb2@gateway/web/cgi-irc/
18:03:06barrywardellI think that still needs to be calibrated in rockbox yet
18:03:19alteregoaso i have to get a ipod 5.5, from wherever
18:03:44scorche|wwell, there are other devices than the ipods
18:03:58alteregoabut not compatible with my car stereo
18:04:28scorche|wthis is assuming rockbox is too
18:04:51alteregoayeah, thats why i have to chose rockbox and ipod
18:04:51 Quit Alexandro5 (Client Quit)
18:05:49 Nick advcomp2019_ is now known as advcomp2019 (n=advcomp2@
18:06:05alteregoacaptain future may help me
18:07:19lostlogicdidn't we used to have two preamp settings for replaygain, one for files with tags and one for files without tags?
18:07:27lostlogicIf not, do we want such a feature?
18:11:13 Join lazka [0] (
18:14:13Lloreanlostlogic: Personally, I'd rather merge the EQ precut and the Replaygain Pre-Amp into one.
18:15:16LloreanThe whole point of replaygain is to give your files a standard level of reference, a pre-amp for non-replaygained files wouldn't do much good simply because they don't have one, I would think, unless you're just fortunate enough to have a collection that's all about level anyway right?
18:16:08 Join roolku [0] (
18:17:16roolkuLlorean: the problem comes when you listen to a mix of files with and without replaygain tags. The ones without are significantly louder.
18:17:19 Join linuxstb [0] (
18:17:26 Quit linuxstb (Remote closed the connection)
18:17:27Lloreanroolku: Then tag all your files...
18:17:45*roolku would like a pre-cut setting that only effects non-replaygain tracks
18:18:21roolkuif you show me how to tag mod, sid and other file formats that are not supported I will
18:19:26 Join ToHellWithGA [0] (
18:19:32LloreanSo, if that's your complaint, why not fix the formats?
18:20:36roolkuwhere did you see a complaint?
18:20:36lostlogicLlorean: It just happens sometimes that you miss a file or forget to tag and then the volume is all messed up −− like this fricking morning when papa roach woke me up with his untagged 5db too loud self at 4am :-P
18:21:11roolkualso I don't understand what you mean by fix?
18:21:12Lloreanroolku: Well, you seem to think it's Rockbox's job to handle the fact that formats don't have sufficient tagging.
18:21:28roolkuit is not broken?
18:21:45LloreanWell, it doesn't meet your needs apparently
18:21:51LloreanSince Rockbox needs to code a fix, or so you feel.
18:22:02lostlogicLlorean: it's just a nice feature to say "If I've tagged the music specifically, use the tags and this preamp, otherwise use this other preamp to avoid hearing loss"
18:22:23roolkuI certainly think that computers should make life more enjoyable and not harder or unpleasant
18:22:25lostlogicI think foobar2k has such a thing
18:22:54LloreanI just think that the solution to every problem isn't "add another option to Rockbox"
18:23:16roolkuI don't see the world in black and white (broken fix)
18:23:20amiconnI thought the preamp was just there to adjust the volume for replaygained tracks
18:23:31Lloreanamiconn: It's cumulative with the EQ precut though.
18:23:38amiconn...and that preamp values is not applied to non-replaygained tracks
18:23:45LloreanI'd rather just have a "Digital Gain" option and be done with it.
18:23:51roolkuI see an opportunity to improve things, and since the topic came up I voiced my opinion
18:24:29amiconnSo this one option should be enough to level the difference between replaygain tagged and non-tagged tracks
18:24:48amiconnMore options is often not an improvement
18:25:00Lloreanamiconn: It certainly can, do +6 on replaygained tracked, then lower overall volume.
18:25:05lostlogicamiconn: hmm, you may be right, perhaps I can use the eq precut and a positive gain on the replaygained tracks.
18:25:12roolkuLlorean: currently they are enabled/diabled with their respective purpose - how do you handle it when you have a combined value and turn one of the features on/off?
18:25:42lostlogicthe problem for me is that the iPod volume doesn't go soft enough −− I use replaygain, -6dBpreamp and lowest volume on my ipod and that's exactly quiet enough for my listening at work.
18:26:03Lloreanroolku: Are the features going to turn themselves off, or is the user going to make a conscious decision? If so, do you think they're intelligent enough to adjust the gain too?
18:26:09amiconnLlorean: I can't really test. I don't have a single track with replaygain tags, and no plans to change that.
18:26:11 Quit petur ("stkov")
18:26:13roolkuamiconn: I would agree, however for some reason it doesn't seem to work as expected. :( I guess I'll have to investigate
18:26:24Lloreanlostlogic: Then why not put in -6 in the equalizer, and then +6 in the REplaygain?
18:26:35Lloreanlostlogic: You can also replaygain your tracks to a higher level than -89 I believe.
18:26:40amiconn(and I don't use software eq either)
18:27:09roolkuLlorean: no need to have a dig at my intelligence :) I am asking for convenince.
18:27:21Lloreanroolku: Convenience is not the only consideration to make
18:27:26LloreanPeople already find our menus too confusing.
18:27:26lostlogicLlorean: I'm going to try that out −− I thought that the preamp was applied to all tracks, not just those with replaygain tags.
18:28:09Lloreanlostlogic: I'd prefer if there was just one "Digital Gain" setting, since having two settings that both do essentially the same things (though both conditional upon another feature) leaves it hard to know what's affecting your music for a lot of users.
18:28:40LloreanPersonally I wouldn't mind at all if the EQ automatically precut an equal amount to whatever band you set highest.
18:28:54lostlogicLlorean: yeah, but a single digital gain feature is insufficient for all use cases as we've just discussed
18:29:10lostlogicbut it sounds like the existing features are sufficient, I just didn't know how to use them.
18:29:13Lloreanlostlogic: Only assuming bad tags, and I'm not sure it's Rockbox's job to correct bad tagging.
18:29:18lostlogicnot bad tag
18:29:33lostlogicit sometimes happens that you get music from a friend and want to listen but don't know whether it's tagged
18:29:39lostlogicor you rip a new CD and forget to tag it or wahtever
18:29:53LloreanWell, then, "missing tags"
18:29:55lostlogicor you just choose to only replaygain part of your colection
18:29:57roolkulostlogic: did it work for you? I can't seem to decrease the sound output for my mod files with the replaingain pre-cut
18:30:03amiconnLlorean: The auto-precut was already discussed a lot, and the consensus is that we do not want it for eq
18:30:20Lloreanroolku: What was said was specifically that replaygain preamp will NOT affect non-replaygain files.
18:30:24Lloreanamiconn: Why not?
18:30:30lostlogicroolku: what I'm going to try (not now, a bit later) is using eq precut to apply to all files and then replaygain preamp up to match the replaygained files
18:30:35 Join atsea-32 [0] (i=atsea-@gateway/tor/x-bde7d80761065d22)
18:31:09lostlogicLlorean: I do agree that getting the replaygain vs. eq settings clarified would be good, but I'm not sure on the right approach at all.
18:31:56Lloreanlostlogic: I'd settle for an automatic EQ precut, and then a "Digital Gain" that could be set to "All, RG Only, Non-RG Only" or some such.
18:32:13amiconnLlorean: If your original source is rather quiet, you don't need to precut as much. If you do, you lose SNR
18:32:18LloreanAnd maybe an option to turn the auto-precut on/off.
18:32:32roolkulostlogic: yes, that is what I had in mind, but I couldn't reduce the volume with the replaygain pre-cut/amp for untagged files the last time I tried
18:33:05lostlogicroolku: the precut is in eq, not replaygain
18:33:06Lloreanroolku: It's been said a few times now, Replaygain preamp will not affect untagged files...
18:33:13*Llorean wonders if he's already on ignore.
18:33:21lostlogicback in 20m
18:34:20Lloreanamiconn: I'd just like the options to seem less redundant. Digital Gain seems like something that's not exclusively used with the EQ (and should not be dependent upon its on/off state)
18:34:58amiconnThe EQ precut is always applied afaik - so that's your digital gain
18:35:03roolkulostlogic: couldn't remember the name, it's the pre-amp in replygain I meant - just different terms for essentially the same thing
18:35:10LloreanYes, but it's not labelled or available as such to must people
18:35:26amiconnAnd replaygain preamp is to adjust the replaygain normalization level vs. untagged files afaiu
18:35:32LloreanYes, it is.
18:35:58amiconnSo it only applies to replaygained files - and you can't keep that functionality with just one digital gain
18:36:05Lloreanroolku: One last time: He said EQ Precut is what you want, NOT replaygain preamp
18:36:10*roolku just cranked up the replaygain to maximum with no change in volume (eq off, replaygain on)
18:36:29amiconnSo we cannot remove one, but we also don't need more than we currently have
18:36:30roolkuthe file in question is a mod
18:36:59 Join Lakshmi [0] (n=chatzill@unaffiliated/lakshmi)
18:37:10amiconnroolku: So it's not replaygain tagged, and rockbox behaviour is correct
18:39:18roolkuI guess I should amplify all my replay gained files to bring them in line with the untaged ones, instead the other way round
18:40:58amiconnYou just need to adjust the replaygain preamp for this while playing a replaygained file
18:43:01 Join bluebrother [0] (i=yEqupppb@rockbox/staff/bluebrother)
18:43:01 Part Lakshmi
18:43:02 Join hcs [0] (n=agashlin@rockbox/contributor/hcs)
18:43:07 Join freqmod_nx [0] (
18:45:59 Part pondlife ("Gone")
18:47:34roolkuamiconnThe EQ precut is always applied afaik - so that's your digital gain <−−- unfortunately not true, eq must be on for it to be applied
18:47:56 Quit lazka (Remote closed the connection)
18:49:11 Quit roolku ()
18:55:12lostlogicgah, roolku still didn't gather that there were two different settings we were talking about :(
18:55:26lostlogicLlorean: the combination of eq precut and replaygain amp gives me what I need.
18:55:38lostlogicI too would like to see all of those settings moved into one menu somehow
18:56:00lostlogicand then name the eq precut and replaygain preamp as two similarly named things that say "This does all files, this only does replaygain files"
18:56:16 Join nomel [0] (i=4782332a@gateway/web/cgi-irc/
18:56:25 Join mirak [0] (
18:57:20nomelon my c200, i notice that ape has pauses in playback between disk reads (as indicated by the icon) this a limitation of the device from read speed or something?
18:57:46lostlogicnomel: during disk reads or between?
18:58:15mirakanyone uses ubuntu here ?
18:59:05GodEaternomel: I would imagine that it's because ape is extremely hard work for the cpu in the player to decode
18:59:12pixelmaI think ape doesn't play too well on portalplayer devices
18:59:20GodEatermirak: this is a rockbox channel - not a linux support channel
18:59:21n1snomel: it's a known bug with high bitrate formats such as flac or ape
18:59:52lostlogicn1s: what's the problem? buffer refill not triggering fast enough?
18:59:54pixelman1s: isn't flac very fast?
19:00:05n1snomel: thry setting the "anti-skip" thingy higher and see if it helps
19:00:06GodEateryeah, flac is fine
19:00:11karashataFLAC works fine on my H10, seeking and everything even...
19:00:18n1slostlogic: seems like it.
19:00:43nomelit plays, then the disk read indicator turns off, there's a delay, the disk read indicators comes back on, then it starts again.
19:00:43mirakGodEater: I need to find someone that uses Ubuntu and it's related to rockbox
19:00:46n1spixelma: yes the decoding is fast but somehow the rebuffering starts too late
19:01:00GodEatermirak: just ask your question then
19:01:10nomelabout 700ms pauses, 2.5 second playback.
19:01:20pixelman1s: I'm not sure and can't take a look atm but I think there is no anti-skip setting on the flash based c200
19:01:30mirakGodEater: I don't manage to encode mp3 with mencoder
19:01:33n1spixelma: oh
19:01:37nomeln1s, i'll try that.
19:01:38lostlogicnomel: sounds like the decoder needs data faster than the buffering system is supplying it because there's kinda a cycle delay between buffer on and buffer off −− like n1s said, increasing the anti-skip will help because it will try to trigger the buffer fill faster.
19:01:41mirakGodEater: for mpegplayer
19:01:49n1slostlogic: if you are interested i reported it back in december
19:01:55GodEatermirak: then that is NOT a rockbox question. You need to go and ask in a ubuntu support channel
19:02:40mirakGodEater: I prefer ask here because it's more likely to find people that uses mencoder for mpegplayer
19:02:50n1snomel: so you have a pause about every 2.5 seconds?
19:03:04mirakGodEater: I asked there too of course
19:03:29 Quit Frazz (Read error: 104 (Connection reset by peer))
19:04:02nomeln1s, yeah.
19:04:11nomelcan't find an anti-skip in general settintgs.
19:04:29n1snomel: then it is not the thing I mentioned as i thought you only had pauses when it rebuffered
19:04:45nomeli think that is what's happening.
19:04:53n1snomel: you are simply using too high compression level for our decoder to handle then
19:05:02 Quit jhulst (Read error: 113 (No route to host))
19:05:20pixelman1s: btw. I had no problems playing flac files with it
19:05:24nomelyou know that because it's playing for 2.5 and only pausing for about .7, meaning it should be able to keep up?
19:05:34scorche|wmirak: if i came in here and was asking how to use a webbrowser, that wouldnt be ok just because i want to browse
19:05:47nomelif it were buffering that is.
19:05:56n1spixelma: I suppose flash targets are not affected as they don't have to wait for disk spinup
19:06:02lostlogicnomel: if you go to system|debug|view audiothread you can tell us which buffer is running low.
19:06:20mirakscorche: yeah right
19:06:48n1snomel: if it is pausing like that all the time then yes the decoder can't keep up
19:07:07scorche|wand if i came in here asking how to rip music from a CD, it wouldnt be ok just because i would be using it on rockbox
19:07:26n1salso it will probably keep buffering because there will not be enough cpu to read very fast from flash
19:07:44nomeln1s, pcm
19:07:54nomelit doesn't even make it to half way.
19:07:59mirakscorche|w: what I want is find people that use ubuntu that was my first question, and it's more likely that here someone might tried to use mpegplayer. Don't do like I said anything else, that's anoying
19:08:22n1snomel: that is the decoder not being able to keep up, go with lower compression or another format
19:08:36scorche|wmirak: my point is that what you are asking sint rockbox related
19:09:02GodEatermirak: and you're required to stay on topic here. So you're in violation of our channel guidelines
19:09:02nomelso what is codec buffer? raw data for the codec buffer?
19:09:21mirakscorche|w: yes it's related because I am trying to encode a video using rockbox wiki.
19:09:24nomelvery cool.
19:09:52mirakGodEater: then remove the howto to encode mpegplayer videos of the rockbox wiki
19:09:58GodEatermirak: the wiki gives advice on how to do it. It's not a supported solution.
19:10:14nomelhow much cpu is dedicated to codec?
19:10:21scorche|wmirak: that is provided for your benefit...if the wiki is wrong, we would sure liek to know how to correct it, but this channel is not for supporting mencoder
19:10:39GodEatermirak: you're expected to show at least a little nouse, and work out if there are differences in the instructions for your distro.
19:10:48n1snomel: it has high priority so it uses as much as it needs
19:11:32 Join hc1 [0] (n=agashlin@nat01-southtowers-ext.Rutgers.EDU)
19:11:33mirakscorche|w: that why I search someone that use ubuntu. if you are not concerned just don't answer, it would made life easier for everybody and avoid goind on useless arguing
19:11:55 Quit hcs (Nick collision from services.)
19:11:57mirakGodEater: well at first I provided some commands in the wiki so ...
19:12:00 Nick hc1 is now known as hcs (n=agashlin@nat01-southtowers-ext.Rutgers.EDU)
19:12:08scorche|wmirak: so we should just ignore any and all rule violations?...what would eb the purpose in having guidelines at all then?
19:12:54GodEatermirak: Llorean told you this was off topic earlier. And you've just ignored him and come back again.
19:13:46Nico_Plostlogic: here?
19:14:04mirakscorche|w GodEater : my first question was trying to find someone that use ubuntu on this channel. You then deviated to this useless arguing about the usefullness of my question. I am looking for people that use Ubuntu and which center of interest is rockbox, and encoding an mp3 with specific options.
19:14:22GodEatermirak: which we KEEP telling you is OFFTOPIC
19:14:33mirakI have better chances to find convergence than asking on #ubuntu if someone uses rockbox
19:14:48*rasher points at #rockbox-community
19:15:56***Saving seen data "./dancer.seen"
19:16:32n1sNico_P: have you done any changes to the low audio buffer watermark in mob?
19:17:01Nico_Pn1s: yes, I've changed how the conf_* variables are set yesterday
19:17:21scorche|wmirak: you deviated from the channel topic by asking an off-topic question...this isnt a rockbox question...this is a mencoder/ubuntu question
19:17:36Nico_Pn1s: that is, I changed it from how it was before in mob... it's closer to svn now
19:17:41scorche|wit even says to take offtopic to rockbox-community if you wish
19:17:44scorche|wend of story
19:18:02n1sNico_P: have you seen FS #6479 ?
19:19:14mirakscorche: like I said it's not a bad place to find people with the interest convergence. and you should have least pointed me to this other channel instead of beeing unsocial.
19:19:50 Join bertrik [0] (
19:19:53mirakscorche|w: so I went on the other channel it's perfect see you there
19:20:01lostlogicNico_P: yeah, I'm around.
19:20:06GodEatermirak: the channel topic points you there#
19:20:10GodEateryou're supposed to read it...
19:20:18scorche|wthat channel is in the topic for this channel, as well as in the guidelines which are supposed to be read before speaking..."interest convergence" is still offtopic
19:20:20Nico_Plostlogic: cool. do you know I'm implementing metadata on buffer?
19:20:22 Join ilgufo [0] (
19:20:35lostlogicNico_P: yes, I gathered that −− how's it goin' ? :)
19:21:01Nico_Plostlogic: quite well... It will soon be ready for commit
19:21:11mirakGodEater scorche|w anyway you should have shown me the offtopic channel just to say me on it I was offtopic there ;-D
19:21:20Nico_Pn1s: the problem is still present in MoB, but maybe it could be extended later
19:21:23zeprobably would have far fewer issues if you'd have made a #rockbox-dev instead of a #rockbox-community :p
19:21:27lostlogicNico_P: I'd love to review it before &| after commit.
19:21:37scorche|w"<scorche|w> that channel is in the topic for this channel, as well as in the guidelines which are supposed to be read before speaking"
19:21:41 Join pondlife [0] (n=Steve@rockbox/developer/pondlife)
19:21:50Nico_Plostlogic: I was going to suggest that :) the code is publicly available already
19:22:07lostlogicI'm sure that I could find it, but you want to link me? :-P
19:22:09 Part pondlife ("Gone")
19:22:16zereally on irc you gotta expect people to show up in the first/most-generic-looking channel they see and then fail to read the topic and guidelines
19:22:19lostlogicah, sweet <3 git.
19:22:31Nico_Pit's on the "mob" branch... you can get a diff and see the log for the branch
19:22:35Nico_Pand many other things
19:23:10Nico_PIt's based around a buffering thread with a new API
19:23:20Nico_Pthat takes some of the complexity out of playback.c
19:23:23scorche|wze: i prefer to expect a bit of respect out of people, instead of blatantly ignoring the topic, onjoin message AND the guidelines
19:23:42GodEaterand then the requests not to ignore the topic afterwards too
19:24:39zescorche|w: good luck with that...can i live in that world too? :p
19:24:43Nico_Plostlogic: the current state is that it works about as well as SVN. Basically I've translated the svn playback code to the new API
19:24:48lostlogicNico_P: a buffering thread independent fromt he audio thread?
19:25:48GodEaterlostlogic: don't go near crossfade with it though ;)
19:25:56lostlogicprobably makes sense, specially with dual core support
19:25:56Nico_Pthe idea is to have the buffering API available for all the threads, so that e.g. the plugins can use it too
19:25:59lostlogicGodEater: hrm?
19:26:14Nico_Pcurrently enabling crossfade in a MoB build fails
19:26:17GodEaterNico_P's code has some issues with crossfade when you turn it on or off currently :)
19:26:29lostlogicI don't use it any way.
19:26:34scorche|wze: so we should adapt to the expectation that no one will listen to us and wont follow rules?...forgive me if i dont...
19:26:42GodEaternor me - I wouldn't have noticed it if pondlife hadn't asked me to try it
19:27:22Nico_Psame here
19:27:38GodEaterwhen you say "fails"
19:27:45GodEaterwhat do you mean too ? I get a hard lock.
19:28:21Nico_Preally? I only got a bunch of codec failure messages... IIRC it was recoverable by simply stopping playback
19:28:29GodEaterI have to reset my ipod
19:28:36Nico_Poh :(
19:28:53GodEaterit's not too onerous - I can manage ;)
19:30:17zescorche|w: all i said was #rockbox-dev would probably end up being much cleaner and on-topic if #rockbox was left to being the community/"front-end" chan
19:30:34zescorche|w: vs the opposite way it is now
19:30:55*Nico_P tries crossfade in the sim
19:31:02 Join stripwax [0] (
19:31:50Nico_Pwoo I get whitenoise
19:32:38 Quit stripwax (Client Quit)
19:36:26 Join ompaul [0] (n=ompaul@gnewsense/friend/ompaul)
19:37:26scorche|wze: but then, we would have more separation of the community (users and developers), which would be undesirable
19:37:54 Quit Rondom (No route to host)
19:39:13preglowyeah, white noise...
19:41:22Nico_PjhMikeS: is it possible to temporarly shut a thread down?
19:42:55nomeln1s, what flac level did you use with your c200?
19:47:03 Quit mirak (Remote closed the connection)
19:49:43 Join Zagor [0] (
19:49:53 Quit Gnu47 ("I'm off to stalk my Pah'Mach'kai")
19:50:11lostlogicNico_P: why would a replacement buffering API impact crossfade? It shouldn't be any different (to the buffering API) than a normal track change...
19:50:48Nico_Pfor me regular crossfade works fine. I think the problem is simply enabling or disabling it during playback
19:51:05lostlogicahh, because that does a buffer resize makemess operation
19:51:37pixelmahe doesn't have a c200, I said that. They were quite high bitrate ones (900...1000 kbp/s), you probably won't have the same problems with flac as you had with ape because the decoder is more efficient by far
19:51:47pixelmanomel: ^
19:52:34Nico_Plostlogic: does crossfading enabling modify filebuf or filebuflen?
19:52:52lostlogicNico_P: yes −− it resizes the PCM buffer and thereby pushes out the filebuf.
19:53:02Nico_PI can't find where
19:53:06 Quit ilgufo ("So Long, and Thanks For All the Fish -")
19:53:07lostlogicat least it did like a year ago when I last looked at it :-P
19:53:52n1snomel: I don't have a c200 but an iriver h300 and i guess I used -8
19:53:54lostlogicaudio_set_crossfade -> line 903 (on your branch)
19:54:05 Join BigBambi [0] (n=Alex@rockbox/staff/BigBambi)
19:54:42Nico_Plostlogic: even in audio_get_buffer I don't see where it changes filebuf and filebuflen
19:56:08 Join petur [0] (n=petur@rockbox/developer/petur)
19:56:08lostlogicNico_P: hmm, this has changed since I looked at it, gimme a few.
19:56:52 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
19:57:57Nico_PGodEater: enabling crossfade just worked quite well on my gigabeat (just a bit of whitenoise)
20:01:57preglowamiconn: you around?
20:02:43 Join jgarvey [0] (
20:04:19nomelmy bad pixelma >_< thanks.
20:04:59hcsThere's a discussion of GPLv3 in #gplv3-meeting shortly, if we want to bring up any of our concerns this might be a good opportunity. Or just for general interest.
20:05:40hcs for anyone who comes across this in the logs
20:06:51BigBambiNico_P: I've been playing a bit more with MoB on the H1x0 and except for the crossfade issue it is working well
20:06:55lostlogicNico_P: the filebuf is acutally set asynchronously because the buffer state is set to trashed by the audio_set_crossfade. See audio_reset_buffer called in SVN playback.c
20:07:15Zagorhcs: are you thinking of any particular concerns that have been raised?
20:07:33hcsZagor: nope, can't remember the discussion
20:09:03 Quit The-Compiler ("Verlassend")
20:10:36Nico_Plostlogic: are you sure audio_get_buffer sets the buffer to "trashed"?
20:10:56Nico_Pin the case of crossfade activation, size != NULL it seems
20:11:17lostlogicI can Nico_P maybe not to trashed, but it somehow indicates to the fill filebuf call that the buffer needs resetting
20:11:57lostlogicat CVS playback.c:2997 it checks if the buffer state is ! INITIALIZED and then calls reset buffer
20:12:15Nico_Pah, true
20:12:35Nico_PI need to reinit the buffering thread
20:12:52 Quit barrywardell (Remote closed the connection)
20:13:04lostlogicyeah, should be easy enough now that you have an api to do so rather than setting a state and waiting for it to be asynchronously picked up ;)
20:13:23preglowmy, playback.c is a monster..
20:13:47lostlogicpreglow: fortunately nico_p's owrk will help tame the monster :)
20:13:53preglowvery good
20:14:11 Quit karashata ("Leaving.")
20:16:42lostlogicNico_P: If I have ideas when I browse your code, should I send thoughts or patches?
20:16:58Nico_Plostlogic: both :)
20:18:40bertrikHi, any AS3514 experts around?
20:23:59 Join mrkiko [0] (
20:24:00mrkikoHi all!
20:24:04mrkikoHi all!
20:24:17bertrikHi mrkiko
20:26:09 Quit jhulst ("Konversation terminated!")
20:26:54bertrikit seems some registers of the AS3514 are never initialised
20:27:24Nico_PGodEater: would you want to try crossfade with a new commit?
20:27:27bertrikthis may have an effect on power consumption and perhaps also on sounds quality I think
20:28:00 Join Dave2 [0] (i=dave@freenode/staff/dave2)
20:28:14rasherbertrik: check svn blame for the driver
20:29:23 Join gstatus [0] (i=4b0b0846@gateway/web/cgi-irc/
20:29:43gstatusi just got a zune today and i was wondering if you can put rockbox onto a microsoft 30 gig zune
20:29:58gstatuscause ive been reading articles on the web and they suggested it is possible
20:30:50n1sgstatus: there is no one currently working on it but it is probably possible
20:31:02pixelmano, Rockbox only runs on the devices that are listed on the project's homepage
20:31:11gstatusyah i figured that
20:31:19n1sI mean possible to _port_ it to the zune it won't work today
20:31:28gstatustoo bad its extremly difficult to mod any type of microsoft firmware
20:31:43gstatusin the begining release stages if you know what i mean
20:32:09gstatuscan i ask you guys an opinion question?
20:32:21n1sgstatus: Rockbox is not a mod of anything and the biggest problem is getting around the encrytpion or signing of the firmware image...
20:32:23 Join criznach [0] (
20:32:49gstatusmy friend is an it tech and he said that in ten years linux and mac will put microsoft out of business
20:33:02gstatusis that a fair statement??
20:33:09Lloreangstatus: This channel is for Rockbox discussion.
20:33:18*n1s cannot see the future but i can see offtopic chatter ;)
20:33:27 Quit gstatus (Client Quit)
20:33:52*mrkiko smiles
20:34:23 Join spiorf [0] (n=spiorf@
20:35:17Lloreanlostlogic: How 'bout moving "Precut" from EQ to its own setting called "Digital Gain" (that always works, even if EQ is disabled), and then rename the Preamp for replaygain to something like "Replaygain Offset" so it's more clear what it's affecting?
20:36:06LloreanThen, if I can convince someone, have the EQ always automatically precut (since this can be offset by digital gain since it's applied at the same step) so that we never get "The EQ is broken, it makes my music sound weird" bug reports.
20:36:14 Join Frazz [0] (
20:37:36preglowman, logf spews junk
20:37:47preglowis there any point in having all these logfs enabled by default?
20:38:20mrkikoWhat logs? Channel logs?
20:38:54Zagorpreglow: I think each logf "producer" should have a #undef logf at the head of the file, to be enabled only during debugging
20:38:57 Quit kugel (Read error: 110 (Connection timed out))
20:39:00preglowZagor: so do i
20:39:15preglowthere's no point in logf spewing junk i'm not interested in
20:40:22ZagorI have disabled tagcache and tagtree while developing usb. otherwise my log lines are drowned
20:40:48Zagori mean logf in tagcache and tagtree
20:41:34preglowplayback.c also has a fair share
20:44:57lostlogicLlorean: waht is this eq problem you speak of? So you just want the eq (when eq is active) to default to -Xdb?
20:45:50Lloreanlostlogic: If I add 6, 3, 2, 2, and 15 db, it should default to -15 for precut. If it's 2, 2, 1, 1, 2 then -2, etc. Then "Digital Gain" can offset this if your music really wouldn't clip, or you're concerned about the lost range.
20:45:52Nico_PZagor: maybe you should commit that change... I get annoyed by the tagcach logf too
20:46:35pregloweach file should just do a macro for related logf events so we can easily disable them
20:46:47Lloreanlostlogic: Increasing gain with the equalizer is a very good way to cause clipping, which then spawns bug reports because people don't know they should counter it with the precut to avoid it. If the precut were automatic at "safe" values, and they could then un-precut with "Digital Gain" I think it'd cause less bug reports
20:46:51lostlogicLlorean: *nod* we also do need to take care that for people who want absolutely no digital alteration to their encoded music we don't do any.
20:46:52LloreanAnd, with proper option naming, be less confusing.
20:47:01preglowLlorean: i'm very much against any automatic adjustment of precut with the eq
20:47:07Lloreanpreglow: Why?
20:47:21LloreanAssuming that you could negate it as easily as you can manually add precut now?
20:47:25preglowi just don't like it, i don't want rockbox to second guess what i do
20:47:37preglowif i apply positive gains, it's because i know the music has enough headroom
20:47:41lostlogicLlorean: also, couldn't people just be smart and keep their eq settings 'centered' aruond zero db???
20:47:47lostlogicI mean that's the 'right' way to use an eq any way
20:47:58LloreanMost of our users have no idea what the "right" way to use an EQ is.
20:48:08mrkikoIf someone has something to say me, then he should send private messages. Bye!
20:48:08preglowpeople should be very used to eqs clipping if you do positive gains
20:48:09 Join kugel [0] (i=kugel@unaffiliated/kugel)
20:48:18LloreanI would hazard that at least half the people with enabled EQs in Rockbox decided that there wasn't enough Bass from the Bass setting, so used the EQ to add more
20:48:33Nico_Ppreglow: I also have a macro to enable logf logging to the debug screen in the sim
20:48:41Nico_Pie to the console actually
20:48:50preglowNico_P: that sounds totally sweet
20:48:57lostlogictheir music is going to sound like crap with the bass cranked up that much any way, so clipping is just the world helping to tell them so :)
20:48:57preglowNico_P: that's where i would expect sim logfs to appear
20:49:14LloreanI think the majority of our users don't know what a precut is. "Rock" "Metal" "Jazz" etc presets don't show you a precut on consumer electronics.
20:49:33scorche|wlostlogic: they dont really care....all other music sounds weak and too diverse ;)
20:49:35preglowLlorean: perhaps we should hard code some presets
20:49:42LloreanSo I'd rather precut be hidden, and give an advanced user a way to un-precut then get bug reports because "the EQ is broken"
20:49:50preglowLlorean: but i'm very much against dumbing down the eq, it's a tool for people who know what they're doing anyway
20:50:09preglowLlorean: i would not expect random user to know what decibels, q and center freqs are anyway
20:50:10Lloreanpreglow: I don't see it as dumbing down. The full range of functionality would still be there, just with slightly different default behaviour.
20:50:23LloreanYou don't need to know what most of those are to use the Simple EQ
20:50:28preglowLlorean: how would i apply a positive gain with no precut using your setup?
20:50:58Lloreanpreglow: If you apply +6 gain in your highest band, then you adjust the digital gain to +6 to counter the -6 "hidden" precut.
20:51:10preglownah, i don't like that at all
20:51:15preglowno eqs i know of work that way
20:51:18preglowall of them work like ours
20:51:24nomelassuming your source is mastered well, you have to apply negative.
20:51:33preglowour treble and bass controls work like you say, if they can, and they're what users who don't know what an eq is should use
20:51:34nomelotherwise you will get clipping.
20:51:48nomelyou can't gain out of the bitrange and expect it to sound good.
20:51:49 Quit atsea-32 (Remote closed the connection)
20:52:03Lloreanpreglow: Maybe automate precut if they select the gain levels in the "Simple" screen?
20:52:07Nico_Ppreglow, Zagor: maybe logf.h should have a bunch of defines like PLAYBACK_LOGF, DATABASE_LOGF...
20:52:08preglowLlorean: perhaps we should make it clearer that the eq screen is for advanced users
20:52:15Lloreanpreglow: I'd go for that too.
20:52:15preglowLlorean: and perhaps hard code some presets like rock, jazz, etc
20:52:21 Quit gromit` (Read error: 104 (Connection reset by peer))
20:52:34nomelhave an outo precut option on by they can do whatever they want. advanced user could shut it off.
20:52:40preglowLlorean: i want to remove the simple eq controls altogether, btw, the graphical screen should be all, but it should be modified to be better, of course
20:52:51Lloreanpreglow: What I'm trying to work out is how to make things more "intuitive". Clearer named options, and less interdependence on options so a user won't *think* they've set something right, get an unexpected result, and be upset because "Feature X isn't working"
20:52:51lostlogicwe could add a prevent clipping option to eq like we have on replay gain.
20:53:07ZagorNico_P: I don't think we need to make a big thing out of it. I'd rather it was disabled by default and then you enabled it on the files you want to debug
20:53:18preglowlostlogic: only kind of "prevent clipping" that exists for eqs is never using positive gains
20:53:43preglowlostlogic: if we rely on automatic precut, the user will be surprised when the volume lowers when analogue volume is at max
20:53:52pixelmapreglow: there are eq presets
20:53:52preglowi want stuff simple, no fancy interdependencies
20:53:53lostlogicpreglow: prevent clipping could do Paul's auto-cut... sounds like preventing clipping to me.
20:54:01preglowpixelma: i know, but they're hidden away, and also files
20:54:16Lloreanpreglow: They could stay files.
20:54:29preglowLlorean: they'd also take minimal space in core, but sure, they could stay files
20:54:33Nico_PZagor: how did you do it? #undef logf in tagcache.c?
20:54:35LloreanBut instead of calling it "Browse EQ Presets" would could simply name it "Equalizer Presets" and be the first item
20:54:43LloreanThen "Advanced Equalizer Settings" could be the second one, and lead to the advance screen.
20:54:58ZagorNico_P: yeah, #undef ROCKBOX_HAS_LOGF
20:55:06Lloreanpreglow: I think just renaming it will be good enough, then users can add more presets easily, and not have to find them in separate places.
20:55:07preglowi don't think i'd want another submenu for eq, i don't much like nested menus
20:55:11 Join gromit` [0] (
20:55:17Lloreanpreglow: It'd actually be less menus, I think
20:55:25ZagorNico_P: but I think the opposite is a better model. i.e. off by default, then add a define in the files you are debugging
20:55:28LloreanOr equal depth
20:56:04*Llorean would really like to remove the on/off option for the equalizer too.
20:56:17lostlogicLlorean: why?
20:56:19Nico_PZagor: all logf calls would need to be changed to a macro, wouldn't they?
20:56:24preglowLlorean: anyway, would you just keep the preset menu as it currently is? no changes apart from position and name?
20:56:27Lloreanlostlogic: You don't turn Bass/Treble on/off.
20:56:37Lloreanpreglow: Yep.
20:56:44lostlogicLlorean: so eq is just off if it's set to zeros?
20:56:44 Join steph_an [0] (
20:56:51Lloreanlostlogic: Exactly
20:56:58ZagorNico_P: you mean like #define logf _logf :)
20:57:01preglowLlorean: if so, i think it'll make presets kind of second-rate. the menu can't remember which preset is active, for example, and the user will still be able to tweak graphical controls like nothing happened, even forgetting a preset is active
20:57:11lostlogicsure fine as lnog as it's done at the GUI level and there's still a single boolean check for it at runtime ;)
20:57:13Zagor(it already is a macro)
20:57:29Lloreanpreglow: Ah, hadn't thought of that. But it'd be nice of users could add their own presets easily...
20:57:31Nico_PZagor: ah... then no problem :)
20:57:39*amiconn would just remove those silly eq presets
20:57:57preglowLlorean: if we are to upgrade presets to something we want users to use more actively, we should make the menu remember if one is active, and also disable the graphical eq screen if one is active
20:58:23Lloreanpreglow: Ah, so require them to choose "Custom" before they can change it?
20:58:35preglowLlorean: would be optimal, if you ask me
20:58:39LloreanSounds good to me.
20:58:42amiconnWhat purpse would the presets be then if you can't fine tune?
20:58:57preglowamiconn: the presets are shit anyway
20:59:07preglowrock? jazz? who cares, it's just for users who have no cleu
20:59:08preglowclue <-
20:59:10amiconnI think th epresets should stay like they are now, or being removed entirely
20:59:34amiconn^ insert 'either' where it fits
20:59:57preglowi agree, but that doesn't solve Llorean's user problem
21:00:14Nico_PZagor: but then I don't quite see what would be in each file's header... if logf is enabled disable it and each dev changes the disabling line?
21:00:18amiconnWhat problem?
21:00:19preglownamely, clueless people use the eq and complain when it sounds crap
21:00:28preglowthey're used to eq functionality being simpler than ours
21:00:46Lloreanamiconn: Mostly the fact that people who use the "EQ" on Rockbox are used to things like the iPod EQ, where there's just choices.
21:00:46preglowbtw, anyone here have a dap with a functioning parametric eq? how does it handle?
21:00:56LloreanAnd when they end up making it clip, they think something's broken and file reports.
21:01:04 Quit petur (Connection timed out)
21:01:09preglowamiconn: i don't think upgrading presets would matter much, even if we put them in the core, they'll eat a minimum of space
21:01:16preglowamiconn: the menu entries are already there
21:01:18LloreanAnyway, my goal is to come up with a bunch of ways to simplify the menu. Surely some of them will suck. :)
21:02:05amiconnWell, an EQ with just presets must not be called EQ imo, even if the actual filtering works like an EQ
21:02:20Lloreanamiconn: It'd still have the advanced mode.
21:02:21preglowLlorean: i think the latest stuff i said balances ease of use and customizability nicely without catering to clueless people too excess
21:02:28Lloreanpreglow: I do too.
21:02:37ZagorNico_P: I'd rather see it being disabled by default so we don't have to add stuff to all files. then each developer adds a line #define ENABLE_LOGF or whatever just before #include "logf.h" in the file(s) he wants to enable logf in
21:02:39amiconnI hate that 'advanced' settings' idea
21:02:46Lloreanamiconn: It's already there in the EQ though
21:02:53amiconnIt hides away settings, by introducing even more settings
21:02:53LloreanWe have a "Simple" and "Advanced" and "Presets"
21:02:59 Join ilgufo [0] (
21:03:07LloreanInstead we'd just have "Equalizer Presets" and "Graphical EQ" as the two options
21:03:10ZagorNico_P: but those defines would never be committed. they're only used during development
21:03:10Nico_PZagor: ha yes that's smarter
21:03:33preglowLlorean: you mean having an "show advanced settings" setting?
21:03:38Nico_PZagor: they could be commented out in SVN
21:04:13 Join WILDROCK_ [0] (
21:04:14Lloreanpreglow: Really, just reorganizing our existing menu. No new options, except maybe having the presets work a little differently
21:04:16amiconnLlorean: 'Simple' and 'Advanced' aren't really simple vs. advanced iiuc - and they just mirror the functionality of the eq screen
21:04:22preglowbut anyway, we should aim to remove the advanced and simple eq menus
21:04:32preglowand instead make the graphical one good for all users, including voice users
21:04:36preglowit's perfectly possible
21:04:36Lloreanamiconn: Simple gives you less options.
21:04:45Lloreanamiconn: Advanced and Graphical are the two "full" sets of options, iirc
21:05:06Lloreaniirc simply just lets you adjust the gain and doesn't concern you with frequency or q at all
21:05:07amiconnI'd drop 'Simple' then
21:05:34preglowthe default state of the grapical screen should be adjusting gain, so that it pretty much serves as a "simple settings" if you don't know about other stuff
21:05:58amiconn..and if the EQ screen can be improved so that it isn't a pita to operate anymore, adjusts to screen size, and is properly voiced, I'd drop 'Advanced' as well
21:06:00preglowand it already is, i think
21:06:08preglowamiconn: that's exactly what i want
21:06:13 Part WILDROCK_
21:06:25 Quit BigBambi (Read error: 104 (Connection reset by peer))
21:06:28preglowamiconn: but it's hard to think of something nice, especially for a gui idiot like me
21:07:12 Join BigBambi [0] (n=alex@rockbox/staff/BigBambi)
21:07:19amiconnAtm the EQ screen controls are far from being intuitive, and e.g. on c200 the 5th band can only be adjusted blindly - it's out of screen
21:07:32preglowcompletely agree, i'd love to see that screen redesigned
21:07:39preglowbut i just have no idea how to
21:07:53preglowi suck at making user interfaces, and i hate coding them
21:08:49LloreanIs it necessary to show all the bands at once?
21:08:55LloreanDon't we just show one at a time on the smaller screens?
21:09:17pixelmaI've been told that this is so on the remote screens
21:09:49preglowamiconn: could you have a quick look at qmf_synth() in to see if i've done some blatant mistake?
21:10:42preglowLlorean: i don't think showing all the bands at once will be very hard, showing all the info about them textually will be, but i don't really think that's necessary
21:11:30Lloreanpreglow: You could just show five sliders, without any text (unlike currently with text for each) then just the currently selected slider's text at the bottom of the screen?
21:11:46amiconnThat might work
21:11:48preglowsome kind of rudimentary design might be five graphical bars, a window header telling you what mode you're in (gain, centre freq, Q/bandwidth), and the value of each param under each slider
21:12:02preglowif there's not enough room for a value under each, show the current one at the bottom of the screen
21:12:06preglowsomething like that
21:12:07preglowthen a mode switch button
21:12:26amiconnShowing vertical sliders would make adjustment more consistent with other options, i.e. more intuitive
21:12:36preglowi want them vertical, yes
21:12:57amiconnAnd at least on colour targets, the actual filter curve could be shown in the background
21:13:09preglowyeah, i've got code for that almost finished
21:13:16preglowjust a couple of bugs to iron out
21:13:23amiconnBtw, presets could be integrated in a similar way as the radio presets
21:13:24preglowand i need a nicer exp/pow
21:13:28pregloweh, log/pow
21:14:11*preglow checks out
21:14:44amiconnI.e. the EQ would stay freely adjustable independently of whether a preset is loaded, and the preset name would be shown in the EQ screen when it matches the current config
21:14:48 Quit ToHellWithGA (Read error: 110 (Connection timed out))
21:15:18LloreanMakes sense.
21:15:19preglowi don't know if that is very valuable functionality
21:15:30preglowif i adjust the eq, i don't care what presets match what i'm doing
21:15:44preglowi'm only interested in going preset -> values, never values -> preset
21:15:53 Join TMM_ [0] (
21:16:00***Saving seen data "./dancer.seen"
21:16:07preglowwhen you're using the radio, it is however interesting to go from value -> preset, to see what station you've happened to stumble upon
21:16:22amiconnIf the eq presets are loaded into an array of struct eq_settings, checking for a match is really just a memcmp()
21:16:50preglowwhich is good, we don't want to spend much more time in the eq than we already do :)
21:16:55preglowcalculation is already pretty expensive on arm
21:17:03amiconnOr rather, as many memcmp()s as presets in the list
21:17:27 Nick TMM_ is now known as TMM (
21:17:42preglowbut i still don't see that as very valuable functionality
21:17:53amiconnThat would require changing the eq preset file format
21:17:55preglowi would never care if my current values match any preset
21:18:09amiconnEven not your own?
21:18:25preglowperhaps :)
21:18:55amiconnFor radio, the .cfg only stores which .fmr is loaded
21:19:30amiconnSo for eq, we could ship an .eqlist with "standard" presets
21:19:30preglowbut this would also mean we can only support a fixed amount of presets
21:19:43amiconnA fixed amount per list, yes
21:19:59preglowi don't know if the list concept is needed for the eq
21:20:15amiconnSomeone who doesn't want /like this list could make his own
21:21:57preglowa builtin list would be cool, though
21:22:14preglowi hate the disk spinning :/
21:22:22preglowthen again, i'll never use it, so...
21:22:26 Join XavierGr [0] (
21:22:30amiconnI think this list concept would be better. Right now, if you want to load one of your own presets, you have to browse through all preinstalled ones
21:22:43 Join donutman25 [0] (n=dagni@
21:23:01amiconnSure, you can delete them - but they'll reappear with the next upgrade
21:23:24amiconnBut with a list concept, you could build your own list and simply ignore the standard list
21:23:38 Join evil-rhino [0] (
21:23:59 Quit steph_an (Read error: 104 (Connection reset by peer))
21:24:05amiconnAnd you wouldn't have the disk spin up everytime, if the eq list is loaded together with the other settings
21:24:57*preglow tries to think of what eq screen lauout would work on something as tiny as the c200
21:25:19evil-rhinohi guys. i've just installed rockbox on a 2nd gen ipod mini. everything works like a charm, except the ipod seems to crash whenever i now plug it in to the usb port on my computer. i get the usb cable picture, but my computer won't recognize it, and the cable picture stays even when i unplug it. the only way to snap the ipod out of it is to reboot. any ideas?
21:26:02evil-rhinoi'm using the current build, fwiw.
21:26:28pixelmaevil-rhino: that happens with some builds and the Mini seems to show that behaviour quite often. You should be able to put it into disk mode manually
21:27:11evil-rhinopixelma: via the apple diagnostic menu, or is there a way to do it within rockbox?
21:28:32pixelmawith pressing the right nuttons - menu+select to reboot and select+play as soon as it boots, if I remember correctly
21:28:43evil-rhinoright, i just did that. will i have to do that every time, do you think?
21:28:56*preglow goes back to his combined assembler and hair-loss fun
21:29:01evil-rhinoit works, btw.
21:29:24pixelmaprobably not everytime but unfortunately I think it won't be the last time...
21:30:25evil-rhinoOh well. not a huge inconvenience. btw, am i now free to delete the standard ipod folders? (iPod_Control and the likes). I don't have any music on the iPod that i wanted to keep.
21:30:33pixelmaand in my experience this is build dependent. So with this particular one it will be everytime
21:30:56evil-rhinooh. about how often do new build come out?
21:31:09 Join MethoS- [0] (
21:31:14pixelmafew times a day :)
21:31:29evil-rhinoohh :) excellent.
21:32:10pixelmathe current build is updated each time something new is committed (except if it's for the manual or rbutil)
21:32:24 Join petur [0] (n=petur@rockbox/developer/petur)
21:32:45 Quit XavierGr_ (Read error: 110 (Connection timed out))
21:32:52 Nick sounddude is now known as CMaster (
21:32:53evil-rhinodo you know about the ipod folders?
21:32:59evil-rhinoi.e. can they be deleted?
21:33:25 Nick CMaster is now known as sounddude (
21:34:05pixelmaif you want to still be able to listen to your music you need to keep the ipod_control etc. folders. I'm not an Ipod expert but I think you don't need them otherwise, someone correct me if I'm wrong
21:34:54LloreanThe ipod_control folder holds any music you used iTunes to put on there.
21:35:06preglowif you don't care about itunes, reatilos or your itunes installed music, they can all go
21:35:09LloreanBut if you aren't using iTunes, you don't need any of the original folders
21:35:14preglowretailos will recreate them all when it starts next
21:35:40evil-rhinoitunes? feh! :)
21:35:49evil-rhinowell guys, thanks for all your help
21:36:00evil-rhinorockbox is pretty, well, rockin'
21:36:37 Quit nicktastic ("Leaving")
21:40:29 Quit desowin ("use linux")
21:40:47pixelmaargh.. I meant "to listen to your music with the apple firmware".... this is when you type to fast :\
21:43:58 Quit donutman25 ("Konversation terminated!")
21:44:12Nico_PZagor: I'm working on adding an ENABLE_LOGF define as you suggested... what about an ENABLE_DEBUGF too?
21:44:38 Quit keanu (Remote closed the connection)
21:45:17ZagorNico_P: yeah I think it should work the same
21:45:39*Nico_P finds the "we open the real file ..." messages very annoying sometimes
21:45:48Nico_PZagor: I'll add that too then
21:46:13amiconnZagor: Regarding USB transfer speed - it might be that even if the controller handles the time critical stuff itself, its speed dram access might depend on the core clock
21:46:27Zagoramiconn: yeah
21:46:45amiconnOh, and of course ata transfers have to be run in parallel, which are done by the cpu uncached memory (!)
21:47:41Zagoryeah but ata is a lot of waiting for the disk, so it's not terribly bus intensive
21:48:45amiconnWaiting for the disk?
21:48:55 Join keanu [0] (n=keanu@unaffiliated/keanu)
21:49:06 Join sarixe [0] (
21:49:25Zagoramiconn: yes, when reading from the disk most of the time spent is waiting for the disk to produce the data
21:49:33amiconnI doubt that
21:50:04amiconnOnce the disk found the sector(s), the bottleneck is most likely our transfer function, not the disk
21:50:09 Join maddler [0] (
21:50:37Zagormaybe. in any case, we'll see how it turns out.
21:51:03amiconnWell, PP OF disk modes vary a lot in speed
21:51:41 Quit gromit` ("Coyote finally caught me")
21:51:51amiconniPod emergency diskmode is very slow on G5, and reportedly on Nano. OF diskmode is also slow on H10 (at least the small one)
21:52:22amiconnOn the other iPods, emergency diskmode is fast (e.g. my Mini G2). The Apple OF diskmode is always fast
21:52:31 Quit Arathis ("Bye, bye")
21:52:37 Quit alteregoa (Read error: 113 (No route to host))
21:53:25Zagoryeah from what I can tell depending on how much effort you are willing to put in the resulting speed can vary a lot. the simplest code will not be very fast.
21:54:04*amiconn wonders whether one of the OFs on PP use DMA for the harddisk
21:54:24amiconnWe don't, as we don't know how (yet?)
21:54:26 Quit sarixe ("(EE) Failed to load "quit" module")
21:54:37ZagorI was just thinking that too. dma would be nice.
21:54:54*Zagor summons Mr.H :)
21:55:48amiconnOn Coldfire it doesn't pay off, as the CPU transfer is faster. It uses line-bursts for memory access. DMA *could* do that, using DMA auto-align, but for some reason that doesn't work for ata
21:55:56*keanu thinks Mr.H is one of the 144 other people in this channel ;)
21:55:58Bagderlike this ?
21:56:50Zagorahh. I'll admit I haven't read that thoroughly yet
21:58:21Zagordefinitely something to play with
22:01:36 Join gromit` [0] (
22:06:01preglowif only this would work :///
22:06:08preglowi hate unfinished stuff
22:06:13 Join donutman25 [0] (n=chatzill@
22:07:08 Quit MethoS- (Read error: 110 (Connection timed out))
22:08:25 Quit nomel ("CGI:IRC")
22:08:35 Join chuklz [0] (n=chatzill@
22:09:21chuklzDoes anybody know of the port to the Gigabeat S and its progress?
22:09:44krazykitthere's a thread in NewPorts, i believe
22:09:51Bagderand a wiki page
22:10:10chuklzbut i didnt know if there has been any more progress
22:10:14chuklzbecause I
22:10:31krazykitif there was more progress, you would see new revisions in the wiki or new posts in the thread.
22:10:40chuklzI'm wanting to change over to linux several times, and the only thing holding me back is gigabeat
22:11:11chuklzokay thanks
22:11:53krazykitchuklz, pm for an offtopic suggestion?
22:12:00 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
22:14:38 Quit chuklz ("ChatZilla [Firefox]")
22:17:48 Join egolost [0] (
22:19:00egolosthmmm.. Is it prefarable to keep the music in the same dir on my sansa when using rockbox as the OF?
22:19:07 Quit amiconn (Nick collision from services.)
22:19:13 Join amiconn [0] (n=jens@rockbox/developer/amiconn)
22:19:20n1segolost: you can have it in any dir you like
22:19:28BagderI don't use the OF...
22:19:52n1sor maybe I didn't understand the question
22:21:11egolostn1s: yeah.. well.. the OF on the sansa seems to have problems with malformated mp3 files.. international characters etc with makes it freeze when updateing the db.
22:21:56n1segolost: I would do what Bagder does then, don't use it ;)
22:22:33egolostSo if i place it somewhere outside the music dir maybe it wont scan those but still rockbox will.
22:22:37BagderNexia NX5850 has hardware decoders...
22:23:26Zagoregolost: it scans all dirs
22:23:27egolostnls: Ill love to not use the OF but theres no firmware support for the USB yet in rockbox.
22:23:32pixelmaegolost: the OF scans all the dirs on my c200. I have no music in their "music" folder and it still scans and also finds them
22:23:57Bagder"Media Interfaces and 8bit RISC MCU of 8051 Compatible"
22:23:58egolostpixelma: ok, then that wont solve it.
22:24:26n1sBagder: that will be a challange to port rockbox to :)
22:24:45BagderI'll smack him with the details ;-)
22:26:28Zagorooh, 8051!
22:26:44Bagderthat nice 90s feeling...
22:29:26Zagorwho has that nexia? some forum guy?
22:29:37Bagderposted to the dev list
22:32:26preglowarghgh, stupid bug :////
22:33:26preglow427% realtime on uwb file
22:33:51 Quit Zagor ("Client exiting")
22:34:07 Join J3TC- [0] (
22:35:23 Quit bluebrother (Read error: 110 (Connection timed out))
22:36:09 Join Zagor [0] (
22:36:50Nico_PZagor: are you sure the default on logf should be to disable? I'm having second thoughts... Maybe a LOGF_DISABLE define would be better
22:37:44preglowholy sheeyat, from 225% to 427% just optimising qmf_synth()!
22:38:28Nico_PZagor: I'm thinking it could be impractical for someone not knowing where a bug is to have to enable logf in all files
22:38:41Nico_Pwhereas we know in which files its output is a pain
22:39:24ZagorNico_P: logf is not useful for finding random bugs anyway. having to disable all output when you want to use logf for development is a pain
22:40:41Nico_Pyeah makes sense
22:41:35Zagoroh man, Debian royally screwed by keyboard
22:41:59Zagorarrow-up is now "screenshot"...
22:42:11*Zagor sighs
22:43:00 Join BigBambi_ [0] (n=alex@rockbox/staff/BigBambi)
22:43:01 Quit BigBambi_ (Client Quit)
22:44:07 Quit Zagor ("Client exiting")
22:45:00 Quit mf0102 ("Verlassend")
22:46:14 Quit Frazz ("Leaving")
22:46:33 Join bluebrother [0] (
22:46:40preglowNico_P: yeah, i agree with zagor here. a random developer never looks at the full logf output to find a bug, he/she wouldn't know enough about the subsystems in question to know what all the output means anyway
22:46:51preglowso it's just garbage
22:47:07Nico_Pthe commit is almost ready
22:47:07preglowin the meanwhile, it's very annoying with all the random output for me when i want to find one of my bugs
22:47:36LloreanI do think you may want full LOGF though?
22:47:57LloreanI seem to recall a few instances where the H100 went into hard freeze, and it was useful to see what the last things in logf were immediately before it.
22:48:40LloreanCould there be a switch/define for enabling all of them, in all files, in just such a situation should it be wanted?
22:48:41preglowthen enable them all
22:48:51Nico_Pthen maybe in that case you could coumment out the part in logf.h that allows fine tuning
22:49:18preglowit would be one #define to comment out per file
22:49:26preglowbut sure, it would be easy for a big switch too
22:49:30Lloreanpreglow: How many files?
22:49:36amiconnWell, if there is a #define LOGF_ENABLE, it should be simple
22:49:59amiconnRight now we have ROCKBOX_HAS_LOGF to enable the logf feature itself
22:50:42 Quit bertrik ("bye")
22:50:44preglow#if defined(TAGCACHE_LOGF) || defined(ALL_LOGF) \ #define TC_LOGF() real_logf \ #else #define TC_LOGF() nothing-here \ #endif
22:50:47LloreanSo then add LOGF_ENABLE_ALL to enable it in all files, otherwise it has to be un-disabled on a file by file basis?
22:50:53Nico_Pamiconn: absence of LOGF_ENABLE overrides ROCKBOX_HAS_LOGF
22:51:00amiconnThe LOGF() macro would need to be changed to only output something if both ROCKBOX_HAS_LOGF and LOGF_ENABLE are defined
22:51:12 Quit gromit` (Read error: 104 (Connection reset by peer))
22:51:23amiconnEach file you want to debug then needs #define LOGF_ENABLE at the top
22:51:36Nico_Pthat's it
22:51:42amiconn...and if you want it globally, just put that define in your config.h
22:52:22preglowamiconn: anyway, what sample rate does our current mp3 voice clips use?
22:53:17preglowamiconn: 16khz good quality speex now decodes at 608% realtime
22:53:23preglowsoon a bit more
22:53:27preglowthis is coldfire
22:53:36 Quit ilgufo ("So Long, and Thanks For All the Fish -")
22:53:43amiconnHow much iram does it need?
22:53:45rasherpreglow: is this voicefile quality?
22:53:50 Nick parafin is now known as parafin|away (
22:53:51 Nick Soap__ is now known as Soap (
22:53:51amiconnAnd what about PP?
22:54:01Nico_Pamiconn: nothing wrong with the #undef / #define ?
22:54:03preglowamiconn: pp won't get this fast, but that's also all i can say
22:54:28 Join gromit` [0] (
22:54:54preglowrasher: and that's cbr, vbr is even better
22:55:01*amiconn now has a really crazy ARM7 vector_add() ...
22:55:29preglowamiconn: the emac unit really, really helped for this one function, and we all know how "good" arm is at tasks like that
22:56:22amiconnYeah, but at least arm (well, arm7) profits from using load/store multiple
22:56:32 Quit J3TC- (Read error: 110 (Connection timed out))
22:56:39preglowamiconn: 16 bit variables :/
22:56:48amiconnThen use some tricks...
22:56:53preglowamiconn: which isn't automatically bad on arm either, but not as easy to use as with emac
22:57:00 Join J3TC- [0] (
22:57:05preglowamiconn: especially since mul and mla doesn't have parallel shifts
22:57:17amiconnThat's what I do in my vector_add, vector_sub and scalarproduct
22:57:23 Join barrywardell [0] (
22:57:44 Join MethoS- [0] (
22:58:18preglowthen i'll have a look at those when i do arm
22:58:21amiconnI know, however, on arm7, it pays off to e.g. load 4 regs (8 values) operand1, 4 regs (8values) operand 2, and then dissect each register into 2x 16bit and do the mla
22:58:45amiconn..rather than using ldrsh for each and every value
22:58:56preglowis ldrsh also 3 cycles?
22:58:59amiconn8 times ldrsh needs 24 cycles
22:59:30amiconnldmia into 4 regs needs 6 cycles, so you have 12 cycles for decomposing until yoU#re not faster anymore
23:00:01amiconnErm, 18 cycles of course
23:00:12amiconnAnd you need 12, i.e. you save 6
23:03:40preglowthis function does 640 outerloop iters with 64 innerloop iters for uwb files, plus another call with 320x64
23:03:45preglowno wonder it got faster
23:03:58preglowthe innerloop does 16xmac
23:06:42 Join Neo [0] (n=chatzill@
23:07:49 Quit Neo (Client Quit)
23:08:05*Bagder is quiet
23:08:11preglowamiconn: do you know what notation gas uses for the shifting mac.x operation?
23:10:03amiconnUmm, no. Never used those yet
23:10:22preglowi'll need it now for free clipping
23:10:26*preglow tries a couple of things
23:11:07 Quit evil-rhino (Read error: 104 (Connection reset by peer))
23:11:44amiconnBut gas understands both motorola and MIT syntax. You can even mix them within one instruction
23:13:14preglowmac.l %rx, %ry, #1/#-1, %accx
23:13:16preglowthat seems to work
23:13:18preglowand do what it should
23:13:39preglowamiconn: gas eats all kind of weird stuff, i accidentally wrote "addl." once, it ate it with no complaints
23:15:03 Join mirak [0] (
23:16:02***Saving seen data "./dancer.seen"
23:27:45 Quit donutman25 (Read error: 110 (Connection timed out))
23:28:54preglowone bit errors
23:32:07 Quit jgarvey ("Leaving")
23:34:21preglowamiconn: the clipping i do is right, right?
23:34:31preglowas per your last pastebin
23:36:25amiconnLooks correct
23:37:33preglowsome place in there there's a one bit error
23:37:43preglowand an extra invert, but that should be easy to find
23:39:39 Quit Rick ("I… don't need to be here.")
23:43:42 Quit J3TC- (Read error: 110 (Connection timed out))
23:44:58 Join Rondom [0] (
23:47:22 Join nomel [0] (i=4782332a@gateway/web/cgi-irc/
23:47:33 Join Rick [0] (
23:48:19 Join Rafagd [0] (
23:48:58nomelany plan for audio dsp plugin support?
23:50:41 Join safetydan [0] (i=dc9d468b@rockbox/developer/safetydan)
23:50:46preglowwhat plugins do you want?
23:50:55preglowsafetydan: feeling up for some audio eq screen refurbishing? :P
23:51:33safetydanpreglow, hah. I was just on the train to work writing down some thoughts on how to redo that screen
23:52:12preglowsafetydan: that's extremely cool, we've had a chat about it here too
23:53:44n1sI defiantly liked the previus suggestions vertical sliders/one screen to rule them all (all the bands that is)
23:53:47preglowthere's some eq chatter from about 20:35 and on
23:53:49preglowa bit gui talk
23:53:58 Join Chuklz [0] (n=Chuklz@
23:54:46ChuklzHello I need some help
23:55:23 Part Chuklz ("Leaving")
23:57:30nomelpreglow: something a little more advanced than crossfeed. something like 4front headphones [] or atsurround processor by adrewlabs [].

Previous day | Next day