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 2007-10-25

00:00:02preglownomel: *shrug*, the plugins part isn't really _that_ relevant, we might as well put it in the core as we do with crossfeed
00:00:15preglownomel: the harder part is finding someone to code it at all
00:00:44 Quit jhulst ("Konversation terminated!")
00:01:16amiconnI would think that such kind of stuff is out of reach for almost all our targets
00:01:29preglowme too
00:01:51preglowdepends how you do it
00:02:12preglowfor iir reverb models, it is possible
00:02:17preglowbut i bet it uses fir filters
00:02:20 Quit spiorf (Read error: 110 (Connection timed out))
00:02:26preglowmight be possible on coldfire, forget it for pp
00:02:50 Quit davina (Remote closed the connection)
00:03:30preglowno, looks like it's iir after all
00:03:45preglowdoing a reverb dsp plugin is on my todo list, but don't expect it soon
00:03:51preglowi code at a snail's pace these days
00:04:36safetydanpreglow, right that's a lot of ideas.
00:06:04safetydanpreglow, how about using some color to indicate potential clipping rather than automatically changing the precut? e.g. the gain sliders turn red when you're in potential clipping land
00:06:35safetydanpreglow, also, there's the Karma EQ if you're looking for other systems with parametric eqs
00:06:44preglowsafetydan: sounds good
00:06:53preglowsafetydan: i'm more looking for people with daps with parametric eqs
00:07:00preglowsafetydan: so they can test how they handle clipping
00:07:29preglowsafetydan: if not change colour, then at least some other visual cue
00:07:51preglowsince colour changes don't come easy on grayscales
00:07:53rasherperhaps a line at the point where clipping would occur (this would also work on bw targets)
00:08:01preglowthat sounds better
00:08:06safetydanrasher, that would work
00:08:13maxkelleypreglow: flashing line on clipping
00:08:33rasherit could still turn red on colour targets of course, but that shouldn't be the only indicator, since there's no warning
00:08:41rasheryou're just suddenly "too high"
00:09:16maxkelleywhen it gets close to clipping, pulse orange?
00:09:24n1ssafetydan: do you mean like when volume+precut+gain for the band > 0 ?
00:09:31LloreanHow do you know where clipping may occur, though?
00:09:35safetydann1s, something like that
00:09:38LloreanI mean, it's theoretically possible with *any* positive gain change.
00:10:14 Quit MethoS- (Read error: 110 (Connection timed out))
00:10:19maxkelleyyou could sort of have something where if it gets close to a threshold several times within a period of time, it'll start to warn..
00:10:25safetydanLlorean, yes. But it at least serves as a warning
00:10:27LloreanWhich would of course mean that the line would be dead center, very easily mistaken for the "center point" line.
00:10:38maxkelleyis a clipping warning really required?
00:10:45Lloreansafetydan: I don't think it'd get interpreted as a warning, honestly. I think it'd be interpreted as "the middle"
00:10:49maxkelleylike, prior to clipping.
00:11:16maxkelleymy poweramps don't let me know before I'm about to clip, they let me know once their relays trigger and the music's off :P
00:11:20n1sLlorean: you should put precut and volume into the equation too
00:11:26Lloreanmaxkelley: I've seen bug reports and forum posts from people adding too much gain, getting clipping, and thinking something is broken.
00:11:43maxkelleyah, ok.
00:11:57Lloreann1s: Volume doesn't really matter, since it's analog and a different question I believe, but precut would of course raise the line, yes.
00:12:07safetydanLlorean, well a warning indicator somewhere on the screen might work. Even if i's a little ! symbol or something in the corner
00:12:13maxkelleyhow about an exclamation mark next to the volume?
00:12:24maxkelleyflashing exclamation.. red on color targets.
00:12:32preglowsafetydan: anyway, i believe the gui should have vertical bar widgets, all fitting in the screen
00:12:42preglowsafetydan: though not necessarily fitting all the settings in text
00:12:46n1sLlorean: I think our dacs do the gain control in the digital domain but I'm not sure, anyone knows?
00:12:55maxkelleyI seem a bit obsessed with flashing tonight :)
00:13:13preglowsafetydan: you would select which bar to work on with left right, and up/down to change value, then a mode switch key
00:13:17preglownot too different from what it is now
00:13:20*Llorean is now a fan of a "prevent clipping" option for the equalizer, enabled by default, but easy to disable for those who actually know what they're doing.
00:13:30maxkelleysoft clipping!
00:13:41preglowLlorean: i think people are bound to be confused by eqs, and that's that
00:13:46Lloreanpreglow: Agreed.
00:13:48maxkelleywe'll be like NAD!
00:13:53preglowLlorean: if we have prevent clipping, people will come asking why the eq controls don't work
00:14:12Lloreanpreglow: Why wouldn't they work?
00:14:39 Join atsea-32 [0] (i=atsea-@gateway/tor/x-f62271d268aa14f1)
00:14:41preglowLlorean: right, i keep forgetting we're not going to compensate for the precut with analogue gain
00:14:56preglowLlorean: in that case they'll just come asking why adjusting gain sometimes lowers their volume
00:15:10preglowpoint is that a parametric eq is complex, not much you can do to help it
00:15:14LloreanWe could *show* the precut in the graphical EQ screen. :)
00:15:19preglowaside from limiting users to presets
00:15:27LloreanThen when "prevent clipping" is enabled, they see it happening.
00:15:28preglowthis is probably why apple goes that route
00:15:45LloreanBetter yet.
00:15:53maxkelleyI wonder if we should have a sort of "Interface Simplicity" level..
00:15:53LloreanShow a "suggested" value on the precut bar.
00:16:01LloreanRather than having any option for it.
00:16:08maxkelleySimple, Intermediate, Advanced
00:16:25LloreanJust a little marker that shows where on the precut bar the "clipping is possible above here" mark is.
00:16:37 Quit BigBambi (Remote closed the connection)
00:16:39LloreanThen you're still leaving everything up to the user, but you're giving them a small visual cue?
00:17:34preglowLlorean: which they almost certainly are not going to understand without reading the manual
00:18:28maxkelleyI was thinking the same.
00:18:30Lloreanpreglow: True, but I've seen eq sliders with the first "bar" being the precut several times. The roaming mark on it will just prompt curious users to actually find out what's going on, and hopefully result in a little education.
00:18:41n1swe should be more like the linux guys, "free software is about freedom (to do really stupid things)" :-)
00:19:08maxkelley"so, if I want to delete something, I do rm -rf /?"
00:19:09n1sbut really, idiot proofing hardly ever works out nicely...
00:19:29preglowi don't think so either
00:19:33maxkelleybecause there are always idiots that don't understand the idiotproofing :)
00:19:34preglowit has to be dead simple, or not at all
00:19:36safetydanokay, so we're going to change the preset handling to be more like the radio stations. Vertical bars. Scale the screen for all targets. Voice. Some use of color. One screen to rule them all and in the darkness bind them.
00:19:58preglowsafetydan: word, but forget about the preset handling for now
00:20:03preglowi'd concentrate on the gui screen first
00:20:04Lloreanpreglow: I think putting the precut on the same screen as the EQ controls isn't a bad idea anyway. At that point, why NOT put a "suggested" indicator on it?
00:20:20safetydanLlorean, yeah everything on one screen will be a goal
00:20:23preglowLlorean: i don't think it's a bad idea either, it just depends on a) space, and b) how well it goes logically with the rest of the screen
00:20:30LloreanI'll give you that then.
00:20:33maxkelleyso, are we talking about the parametric or graphic, right now?
00:20:38n1sOr we could just change the eq to only allow negative gains :-)
00:20:48 Join BigBambi [0] (n=alex@rockbox/staff/BigBambi)
00:20:51safetydaneven if the screen becomes pretty much just text on the smaller lcds
00:20:52Lloreann1s: Nah, that'd be effectively the same as an automated precut.
00:20:57preglowlike, if all eq sliders change mode between gain/q/centerfreq at the touch of a button, how do we show the precut slider didn't change?
00:20:59safetydanmaxkelley, there will only be one screen
00:20:59 Quit ompaul (Client Quit)
00:21:09preglowwe should try to rely on text as little as possible in this screen, there won't be room for much of it on some targets
00:21:14Lloreanpreglow: A horizontal separator between it and the other 5
00:21:16LloreanErr, vertical
00:21:19maxkelleywill it be graphic, though?
00:21:19LloreanSingle pixel wide bar
00:21:30preglowmakes sense
00:21:39n1sLlorean: but it would be a lot less confusing, but i guess the users wouldn't know wht to do when they just want more bass and cant turn up the low shelf
00:21:54LloreanUsually I see them "spaced" to the left of the actual "EQ" sliders, but space may not be free, compared to a single added pixel
00:21:56maxkelleyn1s: there's a bass option in the sound settings.
00:22:11safetydanmaxkelley, that's not enough for some people
00:22:32maxkelleyI'd be fearing for my ears at that point.
00:22:36Lloreann1s: It'd be less confusing, but the idea is not to take away any power from people who are clueful in the process
00:23:07Lloreanmaxkelley: You should see how many people talk about how "great" Rockbox sounds with something like a cumulative 30dB added to the lower frequencies between Bass and EQ.
00:23:22n1smaxkelley: but it's not enough bass!!11!! one
00:24:05maxkelleythose are the people who drive home and wonder why their spoiler walked off the back of their trunk :)
00:24:30n1sLlorean: an eq with 0->-24dB gains combined with preamp would give the exact same possibilities we have now but meh, it won't stop them
00:25:35Lloreann1s: My suggestion was to automate the precut (effectively the same thing) and then someone could un-cut with a "Digital Gain" (preamp) option if they knew they had headroom.
00:25:41n1smaxkelley: usually I guess they use headphones with poor bass response or are just weird ;)
00:25:59LloreanBut it is a bit of a roundabout way
00:26:03 Quit mirak (Remote closed the connection)
00:26:37 Join rotator [0] (n=e@rockbox/developer/rotator)
00:26:44n1sLlorean: i think both these ways are overdoing it anyway, let them loose :-)
00:27:13preglowdoesn't the sansas have decent eqs?
00:27:14n1sbtw we would have to auto-precut volume when it coes over 0 as well as bass/treble too
00:27:32 Join ender [0] (
00:27:52maxkelleyI noticed that the vu meter-like plugins (osc, vu_meter, etc) don't work on radio.. is it possible to get them to work with radio, or do the meter plugins interface with mpd?
00:28:02preglowmaxkelley: it is
00:28:07preglowmaxkelley: afaik, it's on jhMikeS' todo list
00:28:31 Quit ender` (Nick collision from services.)
00:28:33preglowplan is to have radio work with vu_meter, eq, etc
00:28:33n1spreglow: the radio don't pass through dsp at all, no?
00:28:35 Nick ender is now known as ender` (
00:28:40preglown1s: it doesn't pass through the cpu at all
00:28:45maxkelleyah, ok, was just curious.
00:29:01maxkelleyis it routed directly to headphone amp?
00:29:03n1sheh, brain dead raido :-)
00:29:16maxkelleyraid0? on an mp3 player? :P
00:29:26maxkelleyer, sorry, DAP.
00:29:59n1snot yet if I add more disks the h300 will not fit in my pocket anymore :-(
00:30:05maxkelleyheh :)
00:30:42maxkelleyhardcore DAP lovers carry around those rack-mountable raid arrays :)
00:31:24n1sI think it would be better to try to build some kind of adapter for the 160gB 1.8" drives
00:32:32maxkelleyright now, I'm waiting for SDHC support to break past 2 gig :(
00:32:39maxkelleyor, sd at all.
00:33:48preglowamiconn: ouch, clipping seems to heavily affect performance
00:38:32 Join roolku [0] (
00:38:58roolkulostlogic: care to explain what you mean by "gah, roolku still didn't gather that there were two different settings we were talking about :(" ?
00:39:58 Quit CiNc028 ()
00:40:04lostlogicroolku: other than my first statement earlier, nobody was claiming that the replaygain preamp applied to all files
00:40:09lostlogicwhich you seemed to think we were
00:40:19lostlogicwe were saying tha the equalizer precut applied to all files, which it does.
00:40:59lostlogicthat's settinsg|sound settings|equalizer|precut
00:41:07roolkulostlogic: no I wanted a (negative) amplification for files without replaygain
00:41:22lostlogicroolku: yes, I know, that's what I wanted and what I have not
00:41:38gryzorsince i installed rockbox few minutes ago (went well, boots), my computer won't see the sandisk e250 device as a USB mass storage device anymore. What have i missed ?
00:41:41lostlogicI set the preamp in replaygain to +6dB and I set the equalizer precut to -12dB
00:42:52roolkulostlogic: amiconn suggested to use the positive amplification for replaygained files as a workaround, which works, apart from the fact that the lineout on the gigabeat is to high for my stereo
00:43:13lostlogicroolku: yes, so then you use the equalizer precut to bring it back into line.
00:43:28keanugryzor, restart the sansa, and press |<< to boot the OF
00:43:36roolkulostlogic: unfortunately the equilizer precut only works withe the equilizer enabled
00:43:36keanuor restart and connect the usb cable
00:43:51roolkulostlogic: which I would ratert not
00:44:01gryzorkeanu: cable must be connected while starting ?
00:44:12lostlogicroolku: yeah, I agree with that −− I gotta go digging in the code and see how the code paths differ between eq precut and relpaygain preamp
00:44:14keanunot must, but can be
00:44:26lostlogicbecause I'd rather be able to use only one code path for both 'simple' amplifications
00:44:26gryzori dont understand
00:44:43gryzori must used the original firmware to be able to view my device from my computer ?
00:44:50roolkuyep, that would be a good solution
00:45:07rashergryzor: Yes, that's unfortunately necessary, but it's being worked on.
00:45:12gryzoroh ok !
00:45:13 Quit gromit` (Read error: 104 (Connection reset by peer))
00:45:29gryzori read the doc but didnt find this information there :)
00:46:24rasherI'm not sure, but it might be an "information by ommission" - since there isn't information on how to use usb, that means that it's not supported.
00:46:47gryzortypical developer behhaviour :)
00:47:04gryzorbut counter intuitive for newbies, if i may suggest so
00:47:30rasherI agree that it should be mentioned, but with any luck, it might not be necessary for much longer, so I don't know if anyone will bother adding it
00:47:39Lloreanlostlogic: My understanding is that there's only one gain multiply, and it's cumulative of the EQ and replaygain pres if they're both valid for that file
00:48:05Lloreanlostlogic: You may also want to pass on to Roolku in case he still has me on ignore that there's no penalty for enabling the EQ just for the precut, if you've got the bands set at 0 they don't do anything anyway.
00:48:18 Join gromit` [0] (
00:48:44lostlogicroolku: just verified in the code (and with Paul) that the gain is all applied in one place
00:49:05gryzorone more q
00:49:21maxkelleyask away!
00:49:29gryzorrockbox seems to browse only dirs that are named HIGHCASE, lowcase dirs are not displayed
00:49:36gryzoris that intended? or some default option ?
00:50:07maxkelleyerm, you should press-hold on files are select show all files.
00:50:19lostlogicLlorean: but there is the problem that each packet still gets run through that 5 loop to see that there is nothin to do
00:50:31maxkelleyit by default only shows dirs with music or files it can do something with, I think.
00:50:34roolkulostlogic: so it is just a matter of removing the dependency on the eq enable setting?
00:50:39Lloreanlostlogic: I was told there was no performance penalty if a filter was disabled.
00:50:56krazykitgryzor, the MUSIC directory is marked as "hidden" by the sansa's firmware, which is why you can't see it.
00:51:03gryzorahhh, crap
00:51:04LloreanThough it may have been a generalization of "so minimal as to be the same as none"
00:51:12gryzori guess i'll create another then
00:51:15gryzoror display all
00:51:17gryzorty guys
00:51:19 Join Enismirdal [0] (i=411d4f98@gateway/web/cgi-irc/
00:51:20lostlogicLlorean: if eq_enabled is enabled, it enters the eq_process function and runs through a 5 loop doing boolean checks for each packet, even if no channels are enabled
00:51:23 Nick fxb is now known as fxb__ (
00:51:52gryzorapart from these small newbies question, i'm loving rockbox already :)
00:52:06gryzorand i think it's a great project
00:52:25maxkelleygryzor: I felt the same way when I first used it. good to hear :)
00:53:10preglowdoes the h1x0 remote need to be plugged in at startup time to be able to view anything?
00:53:16Lloreanlostlogic: So a case of "minimal, but there"?
00:53:36preglowbah, i'm using a logf build
00:53:38preglowignore me, quickly
00:54:00hcs /ignore preglow
00:54:11Nico_Pargh the buffering debug screen adds 1144 bytes... it's about 90% of the binsize increase from MoB
00:54:18roolkuLlorean: even if there is no performance penalty, if I then want to use the eq (e.g. for headphones) I can't just turn it on (and off respectively) but have to fiddle with presets.
00:54:26preglowNico_P: then don't include it? :P
00:54:47Lloreanroolku: If you have two presets, "off" and "headphones", it takes just as much fiddling as an on/off option.
00:54:51Nico_Ppreglow: I was thinking of maybe removing the audio debug screen instead
00:55:02lostlogicLlorean: yeah, pretty minimal - 1 function call, 15 instructions, 5 of which involve dereferencing to memory.
00:55:03Nico_Pand reducing the buffering debug screen too
00:55:11lostlogicbut that's per packet
00:55:43roolkuLlorean: you forget the disk spinup
00:56:09Lloreanroolku: Are you telling me that you change from your stereo to your headphones so often, and plug in so rarely, that the disk spinup of that will significantly affect your usage pattern?
00:56:37preglowNico_P: well, the pcm buffer level, at least, is nice to have
00:56:38roolkuI agree it is possible to do, but these are all workarounds
00:56:50Nico_Ppreglow: it's in the buffering debug screen too
00:56:53preglowNico_P: i've diagnosed user's codecs problems by having them look at that
00:56:59preglowNico_P: sure, if no info is lost, just axe it
00:57:04Lloreanroolku: Workarounds that don't increase the binary size, or create more confusing options for users, mind you.
00:57:16maxkelleyrockbox should standardize an additional ring on headphone jacks, telling what sort of device is connected by means of resistance values :)
00:57:22Nico_PI'll probably do that... I'm just relieved that the 1k increase is just that
00:57:51*Llorean does not believe in the philosophy "Every idea that anyone can come up with, no matter how trivial, should go in"
00:58:05maxkelleyheh, I know, I'm joking.
00:58:33maxkelleyotherwise, rockbox would be so big that you wouldn't have room to put any music to play with it.
00:59:08 Quit ender` (" It is difficult to produce a television documentary that is both incisive and probing when every twelve minutes one is inter)
01:00:12roolkuLlorean: erm? removing a check could actually decrease filesize?
01:00:36Lloreanroolku: Are you saying that the equalizer doesn't need a precut?
01:01:03Lloreanroolku: If the precut is untied from the equalizer, then it would be impossible for equalizer presets to adjust to the proper digital gain cut without ruining your balanced replaygain as you want it.
01:01:05Nico_Plostlogic: have you looked at my code?
01:01:12Nico_P(I've rebased, btw)
01:01:34lostlogicNico_P: not in any detail, I'll resync and take a look this evening / tomorrow
01:02:00Nico_Pcool. patches are wlecome ;)
01:02:00preglowwhat, we still haven't turned in the tax info? don't we want the money or what?
01:02:02roolkuLlorean: no lostlogic & I were saying it might be good the seperate the pre-amp that is currently tied to the eq on/off setting and apply it always
01:02:06 Quit mrkiko (Remote closed the connection)
01:02:13Nico_Pbut I think I'll commit soon anyway. crossfade seems to be fixed
01:02:14Lloreanroolku: I know what you were saying
01:02:43Lloreanroolku: And I'm saying: The EQ needs a reserved precut so that presets can properly offset boosted frequencies and ensure no clipping.
01:03:01LloreanYour suggestion is to "steal" one feature to use for something else.
01:03:33lostlogicNico_P: that's cool, if you commit then I'll just see it there and can push changes directly ;)
01:03:39LloreanHonestly, if you want your normal audio louder relative to your replaygained files, why don't you normalize them to a level lower than 89?
01:03:48Nico_Plostlogic: indeed. feel free to
01:03:55lostlogicI'm definitely in favor of putting both of those gain settings in a unified place instead of a mile apart int he menus.
01:04:03barrywardellamiconn: around? I've done quite a bit of experimenting with the lcd regs if you're interested?
01:04:13roolkuLlorean: apparently you dont. the presets could still contain a pre-amp setting
01:04:45Lloreanroolku: No, it would no longer be a pre-cut for the EQ. It would be a digital gain setting, independent of the EQ.
01:04:58Lloreanroolku: Meaning that if someone loads an EQ preset, it'll override the perceived separate "digital gain" feature.
01:05:19 Quit n1s ()
01:05:32roolkuI do not understand what you are trying to say with "Honestly, if you want your normal audio louder relative to your replaygained files, why don't you normalize them to a level lower than 89?"
01:05:32LloreanUnless you want to leave it as precut, and separate it functionally only, in which case it's pointless. Again, you can achieve everything you want without doing so.
01:05:40 Quit hcs (Remote closed the connection)
01:06:26Lloreanroolku: Replaygain adjusts your gain values to a specific level. In some scanners you can choose a lower level or higher level, making it quieter or louder relative to your non-scanned audio.
01:06:35roolkuLlorean: "it might be good to seperate the pre-amp that is currently tied to the eq on/off setting and apply it always"
01:07:14Lloreanroolku: Yes, you told me that before. I already read that once. I already told you that I understand what you're saying.
01:07:17roolku"lostlogic: no I wanted a (negative) amplification for files without replaygain"
01:07:35safetydanpreglow, are you going to submit those changes upstream?
01:07:49Lloreanroolku: Mathematically that's the same as positive amplification for files with replaygain combined with a negative amplification for all files.
01:08:31 Quit kugel ("ChatZilla [Firefox]")
01:08:33roolkuLlorean: no you extended it to removing the preamp from the eq presets which I think is not desirable
01:08:56preglowsafetydan: i'm talking with him about it, he's somewhat reluctant to use pure .S files
01:09:00Lloreanroolku: So, you're saying "keep calling it EQ precut"?
01:09:07preglowbut i've decided to keep them that way, so we'll see
01:09:15Lloreanroolku: In that case, won't it confuse users that it WON'T turn off when they turn off the EQ?
01:09:28 Quit rotator ()
01:09:48Lloreanroolku: You can't get away with having the feature at the same time be AND not be "EQ precut"
01:09:49barrywardellamiconn: for the logs: - by "vertically interlaced" i mean this:
01:09:51roolkuLlorean: exactly. but to do this I need to be able to apply negative amplification (without turning on the eq)
01:10:02roolkuwhich I currently can't
01:11:01Lloreanroolku: I don't understand why you can't turn on the EQ, first off. Secondly, I don't understand why you think it's intuitive at all to have an EQ setting NOT turn off when you turn off the EQ.
01:11:10roolkuI don't see a problem. but obviously I can't speak for all users
01:11:40 Quit petur ("Zzzzz")
01:12:13LloreanIt's no longer an EQ Precut if it's not tied to the EQ...
01:12:20LloreanIt's simply a digital gain adjustment
01:13:51LloreanIf I'm using an EQ, and I turn it off, why should I *also* have to go and change the "EQ Preamp" value to 0, when I've already turned it off?
01:14:18 Quit scorche|w ("CGI:IRC")
01:14:53roolkuanyway. In the time we have argued, I could probably have written a patch to do what I want :) But since the topic had come up, I just wanted to voice my support. Bed time now.
01:15:17 Join donutman25 [0] (n=chatzill@
01:15:49LloreanIn fact you even brought up that point yourself, you said you wanted to be able to just turn the EQ off and on, yet your own suggestion prevents this...
01:16:06***Saving seen data "./dancer.seen"
01:17:35roolkuI don't agree, but am to tired to argue any more - good night
01:17:38 Quit roolku ()
01:23:55lostlogichow much slower is if (x&const) versus if (x)?
01:24:19lostlogicis it always one whole instruction, or is it sometimes not?
01:25:14preglowlostlogic: depends, on arm it can be one, one coldfire, never one
01:25:24 Quit barrywardell ()
01:26:31lostlogicpreglow: that's what I thought I remembered. So, I'd like to change the dsp.c so that the check surrounding eq_process is such a int& check, in the interest of avoiding calling that function at all when only precut is enabled but eq_enabled is true
01:26:35lostlogichow do you feel aobut that?
01:27:07preglowwon't save much
01:27:31lostlogica jump and 15 instructiosn with 5 memory accesses per packet in that case
01:27:39lostlogicversus 0-1 instruction in the eq_enabled=0 case
01:27:49preglowit pretty much pales in comparison to just about everything else
01:28:16preglowanyway, a disassembly will tell you for sure :P
01:29:10 Join iamben [0] (
01:37:37 Quit Nico_P (Remote closed the connection)
01:48:17 Join Alonea [0] (
01:48:38nomelis it just me, or does plugging in the usb while playing a song crash the c200 to the point of pulling the battery with the latest build?
01:49:53 Quit Enismirdal ("CGI:IRC (EOF)")
01:54:55pixelmapulling the battery? Never had to do that the worst was a hard power-off (holding power for about 15sec)
01:55:53pixelmabut I know that with some builds the automatic reboot into the OF on usb plug doesn't work correctly (c200 as well)
02:02:07 Join SkinInd95 [0] (
02:03:42 Quit SkinInd95 (Client Quit)
02:05:47 Quit lee-qid ("aufwiederbyebientotsayonara")
02:05:57 Quit Alonea (Connection timed out)
02:06:26 Join makuseru [0] (
02:06:48makuseruis there no build for the current gen ipod nano yet?
02:07:00scorcheit isnt supported
02:07:22scorchelostlogic: you have been talking a dangerous might think you are coming back to us ;)
02:07:26makuseruso theres no way to put music on one in linux?
02:07:51scorchethat is offtopic here
02:08:30 Part makuseru ("Konversation terminated!")
02:09:10nomelhard powerdown works.
02:09:37 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
02:09:38nomelstrange though...didn't do this before. it's when it's playing or not.
02:10:22nomelpulling the batter is faster than waiting 15 seconds :-\
02:10:39pixelmanomel: if that happens and you hold a light on the display can you see a usb plug graphic?
02:12:12pixelmayou can see it shortly before the backlight turns off (for me it freezes on that screen)
02:13:26 Join Bagder_ [0] (
02:13:32 Quit hannesd (Read error: 110 (Connection timed out))
02:15:18nomelhrmm...let me check
02:16:10nomelyes i can.
02:17:00nomeli see the usb logo.
02:18:03pixelmabut I guess that's the same. Observed on Ipods as well and present in some builds, noone's figured out yet why that happens. from last week didn't do it.
02:18:28 Quit nomel ("going home time.")
02:29:10 Join daurn [0] (i=daurn@unaffiliated/daurnimator)
02:29:47 Quit Bagder (Read error: 110 (Connection timed out))
02:29:52 Part pixelma
02:30:24 Join Ebert [0] (
02:31:04Eberthas anyone experienced on Gigabeat the time disappearing and then sometimes reappearing, some hours behind the actual time (well, incorrect time)
02:31:28Eberti set it, boot up, works fine until the next few boots. then there are dashes
02:36:09 Join J3TC- [0] (
02:46:55 Quit daurnimator (Connection timed out)
02:47:05 Join RoC_MasterMind [0] (
02:57:20 Join bb__ [0] (
03:02:13 Join JdGordon [0] (
03:03:59 Join TotallyInfected [0] (
03:04:25 Join SirFunk [0] (
03:05:54 Join Alonea [0] (
03:07:03 Part TotallyInfected
03:07:57 Quit SirFunk (Remote closed the connection)
03:08:32 Quit bb (Read error: 110 (Connection timed out))
03:08:53lostlogicscorche: fear not, I'm still an illusion.
03:10:03lostlogicIf anyone was paying attention to all the argument about eq precut, here is what _I_ want done with it in rough patch form: All this does is eliminate the call to eq_process for each packet when only precut and not filters are enabled. It's ugly and full of magic, but it works.
03:10:28 Join TotallyInfected [0] (
03:11:25 Quit Thundercloud (Remote closed the connection)
03:12:35safetydanlostlogic, so basically use a bitmask instead of an array for whether eq bands are enabled?
03:13:17lostlogicsafetydan: yeah, which is really just a side effect to avoid more code _size_ changes. The real goal was to make eq_enabled (the boolean) only be on if some filters are on
03:16:09***Saving seen data "./dancer.seen"
03:22:25 Join SirFunk [0] (
03:22:45safetydanlostlogic, seems fine to me, but as you say it's a little "magical"
03:24:02lostlogicI started to write macros to make it more readable but didn't feel like it ended up better
03:40:24 Join sarixe [0] (
03:54:56Lloreanlostlogic: I like that approach to, by the way
03:55:15LloreanRather, I like the conceptual approach, haven't read the diff yet
03:55:20 Quit sarixe ("(EE) Failed to load "quit" module")
03:55:54 Join TotallyInfected_ [0] (
03:55:57 Quit TotallyInfected (Read error: 104 (Connection reset by peer))
03:56:06 Quit TotallyInfected_ (Remote closed the connection)
03:56:46 Join TotallyInfected [0] (
04:00:30lostlogicLlorean: thanks, now if only I could conjure up a way to make it less ugly.
04:01:25*JdGordon has a working e200 again :)
04:13:32 Join kubiixx [0] (
04:13:32 Quit kubiix (Read error: 104 (Connection reset by peer))
04:16:40 Quit Alonea (Read error: 104 (Connection reset by peer))
04:19:49 Quit iamben (Read error: 104 (Connection reset by peer))
04:19:56 Quit donutman25 ("ChatZilla [Firefox]")
04:26:30 Join sarixe [0] (
04:29:02 Quit miepchen^schlaf (Read error: 110 (Connection timed out))
04:29:25 Join miepchen^schlaf [0] (
04:43:14 Quit kubiixx (Read error: 110 (Connection timed out))
04:49:19 Part Llorean
04:51:16 Join kfazz [0] (
04:51:46*kfazz discovered something cool today
04:51:54kfazzmy e280 works with my old ipod dock
04:51:55 Quit sarixe ("(EE) Failed to load "quit" module")
04:52:24kfazzsound comes out of the speakers at least, though i never expected the remote to work
04:52:32ATravelingGeekkfazz: I wonder if my e250 will work...
04:52:32kfazzapple protocols and all
04:52:40 Join psycho_maniac [0] (
04:52:48kfazzthing is, it only worked in rockbox, not the OF
04:52:59Eberti find it funny fade on pause doesn't work on lineout, but crossfade does
04:53:29Ebertanyone have the time reset every so often?
04:53:53lostlogicEbert: fade on pause uses the volume control circuitry, so it can't work LO
04:53:59lostlogicEbert: crossfade is in the DSP
04:54:22Ebertah, that's clear. had know-it-all llorean stumped :)
04:54:45safetydankfazz, it's probably just the line out pin being the same between the two
04:54:51kfazzseems like FS #8022 turns off line out for sansa? cuold it be a configureable option perhaps?
04:55:24kfazzwell i saw that the line out pins were the same as ipod on abi forums, but also read the connector pin spacing was different and they wouldn't physically fit, but they did
04:55:35ATravelingGeekkfazz: my e250 doesn't fit in my universal apple dock
04:55:45kfazzmine is an ihome2go
04:55:48ATravelingGeekthe sansa usb cable will fir into the dock
04:56:10ATravelingGeekodd; I guess the fit is just really close then
04:56:48kfazzthen again my sansa is used, so no telling what someone else did to it
04:57:05ATravelingGeekMine is a reburb
04:57:28ATravelingGeekI'll have to try it on someone else's ipod dock/accessory sometime
04:57:40 Quit jhulst (Connection timed out)
05:04:52kfazzEbert , lostlogic: fade on pause works on line out for me, as do the volume controls.
05:05:06kfazzthe ones on the sansa i mean
05:15:21advcomp2019i do not know how you did that because ipod accessories will fry sansas
05:15:42advcomp2019if it goes thru the dock port
05:15:43 Join bb [0] (n=bb@unaffiliated/bb)
05:16:11***Saving seen data "./dancer.seen"
05:17:49 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
05:20:18Soap<Llorean> Personally I wouldn't mind at all if the EQ automatically precut an equal amount to whatever band you set highest.
05:20:30SoapBut if your bands overlap this will not be suficient.
05:21:06SoapWhy dumb-down Rockbox? Removing capability in the name of simplifying seems backwards...
05:24:11psycho_maniacwhat are you talking about?
05:24:41 Join Calcipher [0] (
05:26:40SoapI have the same opinion regarding the conversation you had earlier with lostlogic regarding ReplayGain. It appears common in applications which support ReplayGain to have an adjustable gain/cut for non-RG'd files. If people find a second digital gain entry buried in the menus confusing - they probably should mess with it. One is with the software EQ - and I would assume this is the one most people see. As long as it is clear _in t
05:26:40Soaphe manual_ that the second one with ReplayGain is an /additional/ gain setting (not another entry point to the same gain setting) what is the issue? We expect users to read the manual when it is a question that gets asked 1,000 times, why don't we expect the users to read the manual on this?
05:27:55 Quit bb__ (Read error: 110 (Connection timed out))
05:28:48Calcipherhey Soap, hows things in rb on this day
05:30:28CalcipherI've set up vmware per the wiki suggestions
05:30:39lostlogicSoap: If only the two settings were somewhere in the same proximity (in most audio programs supporting RG the two types of preamp are adjacent)
05:31:17kfazzadvcom2019: the dock is running off batteries, i didn't feel safe plugging it in, but my sansas aren't fried
05:31:56SoapCalcipher: is there a reason you highlighted me? Hopefully not about VMWare, as I don't use it.
05:32:47Calcipherno, when i showed up and sent that, no other users had spoken
05:32:53Calcipherthats all
05:33:07*Mouser_X has spoken!
05:33:33*lostlogic is (as always) an illusion
05:33:40*Calcipher greets Mouser_X
05:34:53advcomp2019kfazz, the ipod and sansa dock pinouts are different
05:37:32kfazzadvcomp2019: i don''t disagree, but my sasna e280 is plugged into my ipod dock, and sound comes out of the speakers, with nothing plugged into the headphone jack
05:37:43kfazzso in one respeck tehy are the same
05:39:01 Quit jhMikeS (Nick collision from services.)
05:39:07 Join jhMikeS [0] (n=jethead7@rockbox/developer/jhMikeS)
05:40:18 Quit RoC_MasterMind ("Leaving")
05:45:55 Part safetydan
05:52:53 Quit Rondom (Connection timed out)
05:53:24 Join Rondom [0] (
05:55:38 Quit RaRe (Read error: 104 (Connection reset by peer))
05:59:10 Join RaRe [0] (
06:00:08 Quit RaRe (Client Quit)
06:02:41 Join ze_ [0] (
06:02:55 Join qweru [0] (
06:10:39 Quit ze (Read error: 110 (Connection timed out))
06:10:40 Nick ze_ is now known as ze (
06:13:28 Quit XavierGr (Nick collision from services.)
06:13:31 Join XavierGr [0] (
06:14:45 Join eigma [0] (n=cat@
06:17:24 Join advcomp2019_ [0] (n=advcomp2@
06:17:38 Quit advcomp2019 (Nick collision from services.)
06:17:40 Nick advcomp2019_ is now known as advcomp2019 (n=advcomp2@
06:37:04 Quit qweru ("moo")
06:41:16 Join RaRe [0] (
06:46:32 Quit XavierGr (Nick collision from services.)
06:46:35 Join XavierGr [0] (
06:51:20 Quit Rafagd ("KVIrc 3.2.4 Anomalies")
07:03:36 Join Mawargen [0] (n=mcn_arge@
07:03:50 Part Mawargen
07:05:56 Quit criznach ("ChatZilla [Firefox]")
07:16:15***Saving seen data "./dancer.seen"
07:30:55 Join advcomp2019_ [0] (n=advcomp2@
07:31:03 Quit advcomp2019 (Nick collision from services.)
07:31:05 Nick advcomp2019_ is now known as advcomp2019 (n=advcomp2@
07:35:45 Nick fxb__ is now known as fxb (
07:39:04 Nick fxb is now known as fxb__ (
07:50:30 Join Asteriskk_ [0] (
07:59:18 Quit eigma ()
08:04:13 Quit Rondom ("Ex-Chat")
08:10:52 Join CaptainSquid [0] (
08:11:04 Quit kfazz ("Leaving")
08:19:52 Quit tedrock (Read error: 113 (No route to host))
08:22:34 Join tedrock [0] (
08:27:03 Quit BigBambi (Remote closed the connection)
08:36:03 Join Rob222241 [0] (
08:39:35 Join norbusan [0] (
08:39:41 Part norbusan
08:45:06 Quit homielowe (Remote closed the connection)
08:49:57 Join webguest11 [0] (i=c03690e5@gateway/web/cgi-irc/
08:50:06 Quit webguest11 (Client Quit)
08:53:16 Quit Rob2222 (Read error: 110 (Connection timed out))
08:54:53 Join ender` [0] (
09:02:45 Quit Mouser_X (Read error: 110 (Connection timed out))
09:11:47 Join Zagor [0] (n=bjorn@rockbox/developer/Zagor)
09:16:16***Saving seen data "./dancer.seen"
09:17:01JdGordonhey Zagor, what do you tihnk about adding a list of warnings which the build table should ignore?
09:19:00ZagorI don't know. are they really that annoying?
09:19:25JdGordonI tihnk they are
09:19:41amiconnBetter don't, imo
09:19:59amiconnOnce they're out of sight, they'll be forgotten
09:20:31amiconnAnd such a filter would most probably also catch real warnings, hiding them
09:21:06JdGordonok, /me drops the suggestion
09:21:51JdGordonfark the mr500 is slow at decoding the jpegs
09:22:03JdGordonabout 7s to do 320x240
09:22:54JdGordon34s to do 1280x960
09:25:22Zagorbig screens aren't always good :)
09:25:38JdGordonim fixing up mpegplayer to compile on it
09:25:47JdGordoncant wait to watch full screen at 2fps :p
09:26:50psycho_maniacwow thats slow
09:27:32JdGordonwe'll find out
09:29:14*JdGordon grabs elephants dream
09:29:39 Nick Bagder_ is now known as Bagder (
09:30:13JdGordonis one of the mirrors in australia?
09:30:32JdGordoni was getting 800+kps!
09:30:37JdGordonkbps even
09:31:00 Join barrywardell [0] (
09:31:12JdGordonhey barrywardell
09:31:26JdGordongot my sansa goign again, and yeah your database refresh fix is working great
09:32:14barrywardellhow did you get the sansa going?
09:32:39JdGordone200tool worked on my laptop after about the 3rd try
09:32:54JdGordonmpegplayer data abort :'(
09:32:57 Join homielowe [0] (
09:34:24barrywardellIt would be nice if e200tool could be made more reliable
09:35:50JdGordonim not sure if its e200tool, or libusb, or linux usb...
09:36:05JdGordone200rinstaller was having the same problem sending the file
09:38:48barrywardellupload_app is what's failing? it's probably libusb or linux usb if it is
09:39:46 Join davina [0] (
09:46:06 Join spiorf [0] (
09:48:35 Join LinusN [0] (i=linus@rockbox/developer/LinusN)
10:02:10 Join Nico_P [0] (n=nicolas@rockbox/developer/NicoP)
10:03:01 Join mrkiko [0] (
10:03:32 Quit psycho_maniac (" HydraIRC -> <- Now with extra fish!")
10:03:44 Nick fxb__ is now known as fxb (
10:03:57 Quit Ebert ()
10:04:02 Join erchino [0] (n=erchino@
10:04:11 Part erchino
10:06:32 Nick fxb is now known as fxb__ (
10:09:22 Join pixelma [0] (i=pixelma@rockbox/staff/pixelma)
10:12:07 Quit Nico_P (Remote closed the connection)
10:14:50 Quit barrywardell ()
10:16:45 Quit TMM (Read error: 110 (Connection timed out))
10:27:50 Quit qwm (Remote closed the connection)
10:28:07 Join qwm [0] (
10:34:24 Join hcs [0] (n=agashlin@rockbox/contributor/hcs)
10:43:09 Join pondlife [0] (n=Steve@rockbox/developer/pondlife)
10:43:44 Join Toki_ [0] (
10:44:21 Quit Toki (Read error: 104 (Connection reset by peer))
10:47:01 Quit TotallyInfected ()
10:50:09 Join barrywardell [0] (
10:50:34pondlifeHmm, my sim locks up if I go into the database info screen...
10:51:42Asteriskk_well, I've just asscertained that the original sansa firmware on the c240 ha no problems reading a micro sd card. I've written to the mailing list about rockbox not reading hte files on that card, and thought I'd try here again see if anybody has more ideas
10:52:26pixelmaAsteriskk_: microSDs are not supported yet (for the c200)
10:52:59Asteriskk_well that would certainly explain it lol
10:53:07Asteriskk_I guess I'll take the card back, then.
10:53:24pixelmaI think that'll come
10:53:36 Join linuxstb [0] (n=chatzill@
10:53:58egolostany updates on the USB stack?
10:54:45Zagoregolost: not really. still working on the basics. currently trying to avoid inducing "babble" while sending large amounts of data.
10:55:14egolostIt's to bad no company could support rockbox instead of developeing their own shitty firmware so you could get money makeing rockbox better.
10:56:10Zagoregolost: well to be honest in reality it wouldn't be only fun having a company using rockbox. they'd have different ideas and priorities, so it would undoubtedly become a fork.
10:56:26Zagorbut I wouldn't mind doing some consulting for them :)
10:56:42Asteriskk_I just wish I knew more about programming so I could fix some of things I want to do with rockbox
10:57:02ZagorAsteriskk_: it's never too late to start learning
10:57:53Asteriskk_the only thing I can do with any reliability is mirc scripting lol
10:58:50Asteriskk_now, that would be a fun idea. control rockbox from mirc
10:59:01egolostZagor: the company could kind of be in the background.
10:59:20egolostZagor: rockbox is strictly c programming.. isn't it?
10:59:26JdGordonc and asm
10:59:32JdGordon99% c though
10:59:40egolostall i know right now is java.
11:00:32Zagorthat's not a bad starting point for learning C
11:02:56 Join webguest07 [0] (i=5b7951c6@gateway/web/cgi-irc/
11:03:44webguest07hi! is USB working on e200 now?
11:03:54JdGordononly in the OF still
11:04:48Asteriskk_so, like, what exactly is the problem with the microsd cards in the c200? is there some sort of complicatioon that needs to be gotten over in order for rockbox to read them?
11:05:30 Join ilgufo [0] (
11:06:34 Quit webguest07 (Client Quit)
11:06:38preglowlinuxstb: did you have a look at the asf tag bug?
11:12:14amiconnlinuxstb, JdGordon: It looks like we have 2 CONFIG_CPU values which actually mean the same cpu??
11:12:33amiconnDM320 and TMS320DSC25
11:13:07amiconnOr am I wrong?
11:13:24Bagderthey're two different flavours
11:13:34BagderTMS320DSC25 and TMS320DM320
11:14:22 Quit parafin|away (Remote closed the connection)
11:14:54 Join parafin|away [0] (
11:15:26 Nick fxb__ is now known as fxb (
11:15:31linuxstbpreglow: Not yet.
11:16:18***Saving seen data "./dancer.seen"
11:17:24amiconnHmm, so the DM320 macro should probably be renamed to make this clear
11:18:23amiconnThe ..DSC25 is arm7tdmi while the ..DM320 is arm926e-j
11:19:16 Part Asteriskk_
11:19:50linuxstbArchopen just calls them DSC25 and DM320
11:22:12amiconnWell, one of the macros should really be renamed, otherwise it's confusing that they both contain a '320'
11:27:12 Join mrkiko_ [0] (
11:31:35 Quit idnar (Nick collision from services.)
11:31:38 Join idnar_ [0] (i=mithrand@unaffiliated/idnar)
11:33:14JdGordonamiconn: yeah, the dm320 folder is tms320dm320 so ill rename the macro to the full name
11:33:59linuxstbamiconn: Done.
11:35:01linuxstbJdGordon: I just renamed TMS320DSC25 to DSC25...
11:35:18JdGordonwe dotn want the full names?
11:35:34JdGordonim happy with the change though.. less work :)
11:35:39JdGordondinner... bbs
11:35:48amiconnlinuxstb: Btw, the PNX0101 isn't as bad as I thought wrt audio processing power. It's an arm720t core (72MHz) - but with an additional dsp
11:36:04linuxstbNot arm7tdmi?
11:38:15linuxstbAccording to this PDF, it's arm7tdmi -
11:39:52 Quit mrkiko (Read error: 110 (Connection timed out))
11:46:10 Join webguest31 [0] (i=d9e1f955@gateway/web/cgi-irc/
11:46:53webguest31Hi, does bookmarking work with .txt files ?
11:47:53 Quit barrywardell (Remote closed the connection)
11:48:21GodEaterwebguest31: no
11:49:55webguest31now theres a great pity, as I have tons of .txt books
11:50:25amiconnAfaik viewer.rock does remember positions in .txt files
11:50:33hcsonly the current position
11:51:16GodEaterand presumably only on one file
11:51:24webguest31yes it does, and I'm grateful for that, but jumping around in a .txt files if difficult without bookmars
11:51:41JdGordoniirc viewer.rock does store multiple bookmarks
11:51:44JdGordonthe last few files
11:52:05webguest31thanks, I'll look into that
11:52:08pixelmaI think I read for the last 10 or so
11:52:23GodEaternot quite the same as bookmarking - but close enough ;)
11:53:03JdGordonstores the last position for as many files as can fit in the buffer...
11:53:20JdGordon#define MAX_BOOKMARKED_FILES (((signed)BUFFER_SIZE/(signed)sizeof(struct bookmarked_file_info))-1)
11:54:22webguest31hmm, but not muliple bookmarks for one file ?
11:55:13JdGordonmanual bookmarks, and multiple for each file would be nice to have though
11:55:37*webguest31 hugs JdGordon
11:55:43JdGordonnow now...
11:56:18webguest31:), thanks for the info
11:56:27 Part webguest31
11:59:33*GodEater polishes his monocole and exclaims "I say!" at all this show of open affection. It's just not British don't you know ?
11:59:48JdGordonI tihnk I'm going to completly rewrite the keyboard for the touchscreen. Should it go in recorder/keyboard.c or create a new folder for it, e.g touchscreen/ ?
12:00:00JdGordonhe's gone GodEater...
12:00:24GodEateryeah I noticed
12:00:36GodEaterI still though it was mildly funny to say though. Perhaps I'm not properly awake yet.
12:01:16JdGordonreminds me of :D
12:01:32JdGordonlink is safe for work ...
12:01:41linuxstbJdGordon: apps/ could probably benefit from a little reorganisation - "player" and "recorder" aren't helpful...
12:02:09rashercharcell/ and bitmap/ on the other hand seems obvious
12:02:16JdGordoneven then its not perfect
12:02:27JdGordonradio shouldnt really be in recorder
12:02:38linuxstbThen maybe just move it to apps/
12:02:54JdGordonapps/ is overcrowded
12:03:23linuxstbYes, but I think radio belongs there. The playback engine could be moved out though.
12:04:40JdGordonwhat about "gui" for ui elements, "screens" for the different screens, that folder is then split into "charcell", "bitmap" and maybe "touchpad", keep "menus" as is and maybe do a bit more cleanup?
12:05:03JdGordonbmp.c shouldnt be in recorder either
12:05:05mrkiko_hey guys! How do I assure the player's playback speed will go to 100?
12:05:13mrkiko_without loosing other settings
12:05:28JdGordonyou cant change the playback speed
12:05:32 Nick mrkiko_ is now known as mrkiko (
12:05:58GodEaterand 100 what? Yaks to the ton?
12:06:01mrkikoYou can
12:06:15GodEaterno you can't
12:06:21mrkiko... or maybe I'm not explaining myself correctly.
12:06:52mrkikoIf I press "play" for two or three seconds, then I have access to a non-spoken menu. Mooving the volume keys here I can change the speed at which the reader reads mp3 files.
12:07:14GodEatermrkiko: which DAP ?
12:07:16JdGordonah, the piutch screen
12:07:33mrkikoDaP = ???
12:08:02GodEaterdigital audio player
12:08:17mrkikoiRiver H340
12:08:21*GodEater notices the time and heads for the airport
12:08:31amiconnThe pitch screen values aren't saved, so a reboot resets it to 100%
12:08:50 Join Arathis [0] (
12:08:54mrkikoAmiconn: wow! Very good
12:09:18amiconnBut there's also a button to reset it to 100% iirc
12:09:30pixelmasomeone knows if battery_bench is working correctly on an e200?
12:09:47mrkikoCan you add somehow this to the list of spoken screens? Maybe there is something useful to change other than speed
12:10:57amiconnAh, yes, it's just SELECT on iriver
12:11:04JdGordonpixelma: it doesnt because the disk is never spinning
12:11:22amiconnmrkiko: You can also reset the speed to 100% by tapping Select (centre button) in that screen
12:12:43pixelmaJdGordon: hmm... I thought I read that this was replaced by something else to make it work (could remember wrong)
12:13:09mrkikoamiconn. thank you so much
12:13:22pixelmaJdGordon: your runtime tests are just based on measuring time yourself?
12:13:57amiconnpixelma, JdGordon: The problem is that battery_bench() does't register an ata callback, but monitors ata activity itself. This needs to be fixed
12:14:14JdGordonif you comment out || ata_disk_is_active() it works
12:15:13 Quit miepchen^schlaf ("Verlassend")
12:15:39pixelmawhy isn't that done officially yet? Is it a not nice enough approach?
12:15:48mrkikoRegarding fs#8003: should I probably test the last firmware? I saw relatively many changes to lists handling...
12:16:46JdGordonpixelma: I agree with amiconn that it should register the callback instead of monitoring disk access itself
12:16:49*JdGordon has a play
12:17:29pixelmamrkiko: you should always check back if the bug still exists in newer builds. Saves a developer from spending time on fixing a bug that doesn't exist anymore...
12:18:01 Quit pondlife ("Read error: 110 (Connection slimed out)")
12:21:29 Quit J3TC- (Read error: 110 (Connection timed out))
12:21:47 Join pondlife [0] (n=Steve@rockbox/developer/pondlife)
12:26:42*JdGordon doesnt think the playback ata callback is used at all
12:27:02*JdGordon have a feeling the disk is shutdown before the playback thread actually acts on the callback
12:28:38JdGordonit definatly isnt used on the sansa's
12:28:53 Quit ilgufo (Client Quit)
12:29:58 Join Nico_P [0] (n=nicolas@rockbox/developer/NicoP)
12:30:59JdGordonit looks fairly useless on the ondio also
12:31:00Nico_Ppondlife: hi
12:31:15pondlifeI'm running the latest
12:31:24Nico_Phow's crossfading?
12:31:45JdGordonhey Nico_P, just the man who can help me :)
12:31:47pondlifeThe wiki is correct AFAIK
12:32:11pondlifei.e. turning it on/off during playback is problematic
12:32:19pondlifeBut the auto-change dir is fixed
12:32:21JdGordonNico_P: are you seeing the ata callback not being acted on?
12:32:31pondlifeI'll let you know
12:32:36pixelmaJdGordon: I remember doing battery_benches on my Ondio
12:32:52Nico_PJdGordon: I'm not using the ata callbacks anymore, but AFAIK they worked well
12:33:10JdGordonwhere you making sure ata_is_active() before buffering?
12:33:28pondlifeJdGordon: Could there be another ata callback, called at the start of spinup, not at the end?
12:33:44pondlifeThat would be ideal for buffering
12:34:00JdGordonI might change it to call on spinup instead
12:34:08JdGordonexcept that would delay the thing doing the spinup
12:34:34pondlifeDo the spinup, then callback?
12:34:50pondlifeOr is that what you mean?
12:35:15JdGordonit would delay the read/write if we did the callbacks as soon as the ata thread noticed it was active
12:36:20pondlifeIsn't that a potential side-effect of the current MoB polling anyway?
12:36:43pondlifeNico_P: Actually, crossfade seems fine now
12:36:50JdGordonyeah, thats why mob should be using the callback
12:36:56Nico_Ppondlife: I'm reassured :)
12:37:10pondlifeI do think using the callback is much better than polling.
12:37:19egolostIs ther being any work done on supporting the iriver E10? Is someone working on it?
12:37:30pondlifeNico_P: What was the problem with the callback triggering a buffer?
12:37:39pondlifeSlightly slower to start?
12:37:44Nico_Ppondlife: it was called too late IMO
12:38:07pondlifeThat shouldn't matter, as it's only a "chance to top up", not the critical "low buffer"..
12:38:32 Quit DataGhost (Nick collision from services.)
12:38:33Nico_Ppondlife: it was also used to buffer the tracks following the first
12:38:35 Join [Sur`DataGhost] [0] (
12:38:45pondlifeBoth situations should trigger buffering to run as fast as possible until the buffer is tull
12:38:59Nico_Pmeaning there was a "gap" of buffering activity between the first track and the following ones
12:39:29pondlifeOnce the buffering thread is off and running, surely it just fills data.
12:39:32Nico_Pbecause for the first track the audio thread requests buffering, but it lets the buffering thread manage the following ones
12:39:49pondlifeTrack boundaries shouln't matter much...
12:40:41pondlifeI'd just see the buffering thread as being kicked into buffering mode, then reading all data it can. And if it doesn't have enough filenames to fill th buffer then it makes a callback to ask for more...
12:41:12pondlifeCertainly you want to read a buffer-full asap to minimise disk spin time.
12:41:29Nico_PI agree, and the callbacks weren't providing that
12:41:37JdGordonhmm... it looks like the callbacks are called every 2s while the disk is spinning but not busy
12:42:16JdGordonexcept the callback code only calls them every 30s
12:42:16Nico_Pcurrently there's a call available to buffering users to request buffering of a particular handle. apart from that the buffering thread manages the level on itself
12:42:38pondlifeSounds good
12:42:41Nico_Ppondlife: and "buffering mode" is when the disk is spinning, hence the polling
12:42:55JdGordonthis code is wrong!
12:43:07pondlifeAh, ok. So it won't necessarily fill all the way if the disk spins down.
12:43:27pondlifeOr will it reset the timer by reading...
12:43:52 Join Frazz [0] (
12:44:11Nico_Pit probably resets the timer by reading, but as soon as filling is complete, ata_sleep() is called
12:45:18pondlifeSo it'll always "trigger itself"?
12:45:27pondlifei.e. until the buffer is full?
12:45:37pondlifeOr would on an idle callback, rather..
12:46:16pixelmawasn't it done so that buffering itself doesn't trigger "activity" and start buffering again so it could end up in an endless buffer loop? (I had that for a short time on my Ondio)
12:46:48pondlifeThat's what I was thinking of. It needs to be buffering under-the-radar, kind of.
12:47:00 Quit dionoea (Remote closed the connection)
12:47:06 Join dionoea [0] (
12:47:49*Nico_P is a bit lost
12:48:32Nico_Ppixelma: the enless loop prevention is simply done by not using the ata idls callback on targets with less than 8MB
12:48:49Nico_P(in SVN)
12:51:41JdGordonwhy is there a ata_spin() at the end of the wps button loop?
12:53:33Nico_Ppondlife: so is the crossfade issue gone?
12:55:45JdGordonok, does this make sense... if ata_spin() is called in a callback, the disk will stay spinning for an extra few seconds to give threads a chance to work on the disk, but the callbacks will not be called again untill the next time the disk is spun up (i.e after then next spindown) ?
12:55:46pondlifeYes, I think so
12:55:52pondlife^ Nico_P
12:56:26Nico_Pcool :)
12:56:47pondlifeJdGordon: Callbacks should be once per spindown, on the way out, right?
12:57:06 Join micols [0] (
12:57:14pondlifeDoes anything apart from playback buffering use the disk callback?
12:57:15JdGordonsomewhere in the middle of the spin, but yeah
12:57:18micolsanyone having touble with rockbox on nano 1g (4gb model) ?
12:57:21JdGordonconfig saving
12:57:26pondlifeAh, yes
12:57:27JdGordonbut thats not in a seperate thread
12:57:30preglowmicols: what kind of trouble?
12:57:33micolsit lags on playback of mp3s and midis
12:57:40micolshigh pitched noises
12:57:45micolsand midi says something about buffer
12:57:45preglowno, i don't get that
12:58:12micolsi checked System and debug information
12:58:26micolsthe pcm or whatever is at top gets filled
12:58:32micolsthen the pitch noises sound
12:58:46micolsbuffer is always half filled or so (the one at the bottom of debug screen)
12:58:53pixelmamidiplay gives that on all portalplayer devices because decoding can't keep yet
12:59:09micolsyep ive read that midi still is unstable
12:59:12micolsbut mp3 should work
12:59:20micolsflac is even worse.. it plays for 1 second
12:59:30micolsi got fat32 fs on it
12:59:49micolsflac shouldnt be working either i know.. but mp3 should :-)
13:00:30Nico_Ppondlife: I realised yesterday evening that the 1k binsize increase from MoB is mostly due to the debugging screen :)
13:01:06pixelmamicols: flac should work fine, (mp3) too. I could imagine that you have one of the problematic Nanos.
13:01:07amiconnmicols: Flac should work very well. It's a very efficient decoder
13:03:10micolsokay.. it just stops and skips to my mp3s instead
13:03:20micolsand it got high pitched noises even more than the mp3s
13:03:43pregloware you using the eq?
13:03:58micolsyes i tried using the eq at first then i thought it was the one causing trouble
13:04:08micolsi tried stereo separation and bass reducing by -6db
13:04:12micolsstereosep 140%
13:04:20micolsbut then i did reset it to 100% and 0db
13:04:28micolsstill got the trouble
13:04:43 Quit mrkiko (Remote closed the connection)
13:04:44preglowpixelma: do problematic nanos make high-pitched noises, though?
13:04:57micolsone more thing, my cpu speed skips between 30 and 60mhz all the time!
13:05:04preglowi thought micols 80, you mean
13:05:05micolsthat might be the cause
13:05:09micolsyes 40 and 80
13:05:13preglowmicols: that's normal
13:05:23pixelmapreglow: there were reports of faulty mp3 playback at least as one of the symptoms
13:05:24micolsisnt that making it too slow to play the mp3?
13:05:36micolsor at least too slow to play flac
13:05:42micolsi guess it needs full speed for that
13:05:49preglowmicols: nah, we boost the frequency temporarily when we need more juice
13:05:52pondlifeNico_P: Could you combine the debugging screen with the audio one? Or make it debug builds only?
13:05:53micolsthe funny thing is that movie playback is fluent..
13:05:54preglowmicols: flac is waay faster then mp3
13:06:03micolsmpgs with sound seems to be fluent
13:06:08Nico_Ppondlife: I'll probably combine the two screens
13:06:19pondlifeTo be honest, the debug menu could be debug builds only, for a bigger saving :)
13:06:25 Join chelli [0] (
13:06:25pondlifeKeep Out!
13:06:31preglowpondlife: the debug menu is quite nice to have from time to time
13:06:32Nico_PI'm not sure about the number of lines though
13:06:39preglowpondlife: making users diagnoze their own problems
13:06:47 Part chelli ("Client exiting")
13:06:47micolsthe pcm bar at top of debug menu goes all the way up
13:06:55*preglow smells his tea cup
13:06:57micolsisnt that the cause of these high pitched noises
13:07:05micolsit happens everytime i get those pitches
13:07:09micolsit gets filled
13:07:16preglowmicols: no, that's a good sign, it means the decoder is on top of things
13:07:25preglowmicols: eh? the noises occur _consistently_ when the bar is at the top?
13:07:35micolsthe noises seem to be random
13:07:42micolsbut same kind of high pitched noises
13:07:46micolsat random time
13:07:49 Join TMM [0] (
13:08:00micolssometimes it plays without them, but its rare
13:08:11pixelmamicols: it would be easier to read if you could fit more of your thoughts on one line (and would decrease channel log length) ;)
13:08:12preglowi think you have one of those troublesome nanos, yes
13:08:12micolsi thought it could be because the fadein/out plugin
13:08:34preglowa percentage of nanos misbehave with rockbox
13:08:37micolsi tried pausing/playing a few times and i think i got it playing without errors one time
13:08:38preglowand we don't really know why yet
13:08:49preglowwe _think_ it might come from overheating
13:09:06micolsit gets hot indeed.. hotter than with apple firmware
13:09:18micolsbut battery still seems to last long :)
13:09:31pixelmamicols: there's a thread about it in the forums
13:09:39micolscool ill check it out thanks
13:09:49 Join RaZorbacK [0] (
13:13:35 Quit hcs (Read error: 110 (Connection timed out))
13:15:10 Join hcs [0] (n=agashlin@rockbox/contributor/hcs)
13:15:26preglowamiconn: decoding goes from 410% realtime to 290% realtime on one file with heavy clipping, doesn't that seem a bit much to you?
13:16:21***Saving seen data "./dancer.seen"
13:16:40 Quit RaZorbacK ("MegaIRC v3.90")
13:20:39 Nick [Sur`DataGhost] is now known as DataGhost (
13:28:22 Quit linuxstb (SendQ exceeded)
13:28:41 Quit Calcipher (Read error: 104 (Connection reset by peer))
13:35:28 Join Rondom [0] (
13:38:12 Quit spiorf (Read error: 110 (Connection timed out))
13:38:55 Join spiorf [0] (
13:42:48 Part pixelma
13:49:37 Quit pondlife ("Read error: 110 (Connection slimed out)")
14:00:15 Quit SirFunk (Read error: 110 (Connection timed out))
14:02:00 Join J3TC- [0] (
14:07:25 Quit Arathis ("Bye, bye")
14:10:08 Join Alexandro5 [0] (i=4fb45967@gateway/web/cgi-irc/
14:10:52 Join desowin [0] (
14:11:47Alexandro5can someone help me?, where i can find firmware update for SMART SM MP4W
14:15:09 Join petur [0] (n=petur@rockbox/developer/petur)
14:17:48 Join Thundercloud [0] (
14:20:23 Quit Alexandro5 ("CGI:IRC")
14:22:51amiconnpreglow: That must happen in another decoding stage then; the [-32767..32767] clipping in asm cannot be the cause
14:26:26preglowamiconn: perhaps
14:28:14amiconnBtw, what does gas encode is you use msac with parallel load?
14:28:29 Join nenolod [0] (n=nenolod@atheme/developer/nenolod)
14:29:25amiconnAnd: Using mac with parallel load + negation is slower than using msac without parallel load (which is not broken) and load using movem.l
14:29:39preglowamiconn: it's not it... disregard what i said
14:29:55preglowamiconn: it encodes msac.w with parallel load and completely wrong registers
14:30:02 Join kubiix [0] (
14:30:47*amiconn has to try this...
14:30:59preglowamiconn: remove the neg and replace mac.w with msac.w
14:31:01preglowthat's all you need to do
14:31:02amiconnYou're sure you didn't forget a % somewhere, or put one too much?
14:31:11preglowamiconn: just try, it's really fast
14:31:23preglowi'm 99% certain i've done all the right things
14:32:59J3TC-Does viperman3 at iaudiophile forums hang here?
14:33:11preglowamiconn: anyway, i'd rather fix binutils and commit the right thing
14:33:32ZagorJ3TC-: not with that nick, anyway
14:33:48amiconnpreglow: lines 77.80?
14:33:50J3TC-Ah ok
14:34:30preglowamiconn: aye
14:34:31J3TC-I'm wondering if he could share his modded patches such as 2954_scroll-margins_20071019-MOD.patch
14:34:43preglowamiconn: also the four below
14:35:00preglowfor a complete of eight mac.w in order_8 and ten in order_10
14:35:21ZagorJ3TC-: you mean this?
14:36:03 Join RaZorbacK [0] (
14:36:16preglowi hate mixed c and inline asm
14:36:41J3TC-Ah excellent, it got synced :3
14:37:42J3TC-But he has also other patches modded such as progressbar slider and that hasn't been updated
14:39:06Zagorkindly ask him to post them on the tracker. that's what it's there for.
14:39:32J3TC-Ok :3
14:43:13 Quit RaZorbacK ("MegaIRC v3.90")
14:43:36 Join RaZorbacK [0] (
14:47:48 Join MethoS- [0] (
14:57:00*preglow bbl
14:57:50 Join ackbahr [0] (
15:02:36Zagorlovely formatting on the latest list posting...
15:04:28 Join midgey|web [0] (i=8dd3a84d@gateway/web/cgi-irc/
15:06:46 Join nicktastic [0] (n=nick@unaffiliated/nicktastic)
15:07:00 Join barrywardell [0] (
15:07:51markunZagor: looks great :)
15:16:22***Saving seen data "./dancer.seen"
15:20:52Zagormuh. this is so frustrating.
15:21:46Zagor64-byte packets work IN+OUT. 512 byte packets work IN, but not OUT.
15:22:01 Part LinusN
15:23:05markunZagor: what kind of packet sizer are there?
15:24:07markunany luck with the other 3?
15:24:21Zagorno, 32 doesn't work either way for some reason.
15:25:09ZagorI still haven't fully worked out the relationship btwn the request size I'm sending the controller and the packet sizes on the bus. in theory it shouldn't matter how much data I give the controller, but in reality it does...
15:26:12Zagorwhen I send 64-byte chunks to the controller, I get the above 64/512 result. if I send 512-byte chunks to the controller, nothing works.
15:27:06ZagorI'm obviously doing something wrong, but I can't figure out what.
15:28:18Zagorand again the ugly suspicion that I'm not having the right docs is rearing its' head...
15:35:59 Quit ackbahr (Read error: 104 (Connection reset by peer))
15:36:16markunyou are using the i.mx31 docs, right?
15:36:17 Quit CaptainSquid ("Miranda IM!")
15:36:40markunif you buy a Gigabeat S you will know for sure that you have the right docs :)
15:37:03markunperhaps the Gigabeat S guys can run some test for you?
15:37:19markunto know if it's a problem with the doc or something you do wrong
15:38:08markunreceived my 1200mAh battery for my Gigabeat today. Charging is taking forever, but I guess that's a good thing :)
15:38:17ZagorI don't know how such a test would work. I don't know if this is the right docs, and if it isn't I don't know what I'm missing. and that's rather difficult to write a test for.
15:38:25zichonice, where'd you get it?
15:38:36markunipod 4g battery
15:38:48zichoOh, coolio.
15:39:07zichoHow did you change the battery?
15:40:13markunbut I didn't do any soldering like the others
15:40:22Zagorhmm, I guess reading out the id, version and params registers on a gigabeat s might be interesting
15:40:43Zagorany S owners around?
15:44:55Zagorfor S owners reading the log, the contents of memory address 0x43F88000, 0x43F88120 and 0x43F88124 would be interesting. all 32-bit words.
15:46:34 Quit MethoS- (Read error: 110 (Connection timed out))
15:47:44 Quit desowin ("use linux")
15:48:31markunZagor: I'll also try to remind any of them
15:50:22 Quit hcs ("Leaving.")
15:52:13barrywardellZagor: would it be helpful to test on another PP target? I have a e280 and h10...
15:52:45Nico_PZagor: I'm an S owner but I run linux and there's now way to update the firmware from there
15:53:01amiconnpreglow: I think I found the bug in the gas sources - and it's still in 2.17 ...
15:53:24 Nick parafin|away is now known as parafin (
15:54:16Zagorbarrywardell: the readouts of the registers could be interesting, just to see if they're the same or not.
15:54:45barrywardellwhich ones? the whole lot?
15:54:50 Join ackbahr [0] (
15:54:58Zagorthe register offsets are 0, 0x120 and 0x124. pp5020 base is 0xc5000000, pp5002 base is 0xc0005000
15:55:39amiconnThe USB controller in the 5002 is most likely quite different - and we don't need to handle it
15:55:57amiconnThe 3rd gen does *not* use the PP5002's internal USB controller
15:56:03Zagoramiconn: they have a bridge?
15:56:16 Quit Isolinear ()
15:56:22Zagorok, ignore that then
15:56:43amiconnThe 5002's internal controller is USB1.1 only, but the 3rd gen has USB2.0
15:56:55amiconnSo they must have another chip - which we don't know
15:57:35amiconn(unless I am missing something very essential)
15:58:00 Join MethoS- [0] (
16:00:53Zagorhow do we know the 5002 only supports 1.1? some product sheet?
16:02:12Zagorgotta go
16:02:12 Quit Zagor ("Client exiting")
16:02:51 Join scorche|w [0] (n=8dc5049d@rockbox/administrator/scorche)
16:06:14 Join desowin [0] (
16:06:37 Quit ackbahr (Remote closed the connection)
16:09:13barrywardellZagor: for the logs - on my Sansa: ID=0x22fa05, DCIVERSION=1,DCCPARAMS 0x183; on H10: ID=22fa05, DCIVERSION=1, DCCPARAMS=0x183
16:11:41 Join pepie34 [0] (
16:13:38 Quit J3TC- (Read error: 110 (Connection timed out))
16:18:39 Join pixelma [0] (i=d5494daf@rockbox/staff/pixelma)
16:23:34 Join J3TC- [0] (
16:23:45 Join pondlife [0] (n=Steve@rockbox/developer/pondlife)
16:27:26Nico_Plostlogic: ping
16:29:18Nico_Phave you had time to go over the mob code?
16:29:32lostlogicam actually looking at it now
16:29:36lostlogicyou about to commit?
16:29:49Nico_Pnot in the next few minutes but maybe today
16:29:56 Join Llorean [0] (
16:29:59 Quit MethoS- (Read error: 110 (Connection timed out))
16:31:12 Quit pepie34 ("Ex-Chat")
16:33:19lostlogicNico_P: what is the binary size impact of the patch?
16:34:09Nico_Pthere's an aout 1k increase due to the debugging screen. I plan on combining the audio and buffering debug screen
16:35:20 Join Mouser_X [0] (n=someone@
16:36:55lostlogicNico_P: have you done any performance comparisons between the old code and new? Shouldn't really be any difference I guess since the hard storage should always be the buffering bottleneck
16:37:21Nico_Pwhat kind of performance? time to fill the buffer?
16:37:36pondlifeDisk spin time and buffer size I guess?
16:37:37lostlogicyeah, battery performance, playing performance while buffering, etc.
16:38:13pondlifeOn a slow target in particular, not a Gigabeat ;)
16:38:49Nico_Pno I haven't done any runtime tests
16:39:05LloreanOne of the slower buffering targets is the iPod 5.5Gs I believe.
16:39:20pondlifeGodEater_: around?
16:39:27LloreanI seem to recall people reporting what sounded like really, really long times with the disk spun up
16:39:51pondlifeSome people with 64MB RAM reported minutes of disk activity
16:39:57 Join Isolinear [0] (
16:39:59 Nick fxb is now known as fxb__ (
16:40:27lostlogicyeah, the combination of 64M ram plus the slower ATA driver (not sure on the details here, only amiconn would be) would make the 5.5g have LONG spin time
16:40:58pixelmapondlife: if I recall, GodEater is on a plane to Canada
16:41:05amiconnATA isn't any slower than on regular G5 when reading large chunks
16:41:30amiconnThe only thing that is much slower is writing 512 byte chunks
16:41:39lostlogicamiconn: thanks!
16:42:04LloreanWell it may be that both G5 and G5.5 have slower buffering time then.
16:42:17 Nick advcomp2019 is now known as advcomp2019_ (n=advcomp2@
16:42:17LloreanDon't they just have a higher CPU load than average because of the massive screen updates in the WPS?
16:42:34amiconnReading 512 byte chunks (e.g. fat sectors) is a bit slower due to an extra memcpy
16:45:10 Quit Frazz ("Leaving")
16:46:18 Part advcomp2019_ ("Leaving")
16:49:51pondlifeNico_P: I'm still very happy with the way I can skip back to a buffered track with no spinup...
16:50:21Nico_Ppondlife: :)
16:50:37Nico_Pany problems left that should prevent me from committing?
16:51:11pondlifeI've not tested much today, sadly
16:51:50 Quit midgey|web ("CGI:IRC (EOF)")
16:52:00 Quit pixelma ("laters...")
16:52:48pondlifeAnyone able to test with a 64MB iPod Video?
16:54:01pondlifeNico_P: See if you can get the debugging screens merged and the binsize down a bit
16:54:36Nico_PAt one point I was actually thinking of putting the debug screen in an #if 0
16:54:57pondlifeMight be useful to keep it in for the first week at least.
16:55:34pondlifeI'd have the PCM, real, alloc sliders and all the rest just as numbers (all on one screen)
16:56:09pondlifeSorry, I meant PCM, real, usefl
16:56:11pondlifeNot alloc
16:56:35pondlifeWhy does alloc give a slightly smaller value than the entire buffer?
16:57:10pondlife25,156,564/25,163,888 for example
16:57:49 Join jgarvey [0] (
16:57:50Nico_Pprobably because alloc doesn't take the handle struct into account
16:58:30 Join kugel [0] (i=kugel@unaffiliated/kugel)
16:58:31pondlifeThat's quite a lot of memory used for the linked list 7k
16:58:36Nico_PIt's a big difference, though... how many handles do you have?
16:59:14Nico_Pand is data rem zero?
16:59:57Nico_Psizeof(struct memory_handle) is 304 on my sim
17:00:01 Nick idnar_ is now known as idnar (i=mithrand@unaffiliated/idnar)
17:00:15pondlifedata_rem is 330150
17:00:36 Join advcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
17:00:36pondlifeWhat does that represent?
17:00:48Nico_Pthen some of the alloc'd space could be unused... also some data types can't wrap
17:01:08Nico_Pit's the amount of data left to buffer from the files in the buffer
17:01:20Nico_Psome of the files aren't complete
17:01:29pondlifeThe last one is unlikely to be
17:11:48 Join mf0102 [0] (n=michi@
17:14:35 Join Calcipher [0] (
17:15:23pondlifeNico_P: One oddity; my audio debug screen reckons I have 3 tracks, when I only have 2 in the album/playlist. Buffering screen gives the right figure though...
17:16:10Nico_Ppondlife: yeah, both screens use the same calculation but I subtract one to the result in the buffering screen. that's why you can have -1 on startup
17:16:23***Saving seen data "./dancer.seen"
17:18:01pondlifeHmm, why do you need to do that?
17:18:18 Join n1s [0] (
17:18:34pondlifeIt shouldn't be possible to have -1 tracks buffered
17:20:54Nico_Pit's a weirdness in audio_track_count() that I didn't even bother checking
17:23:32 Nick fxb__ is now known as fxb (
17:23:48 Join bluey- [0] (
17:25:11Nico_Pnow to merge the debug screens...
17:25:16 Quit J3TC- (Read error: 104 (Connection reset by peer))
17:25:31Nico_Ppondlife: how much space does the current buffering debug screen take on your H300?
17:25:48Nico_PI'm not sure how many lines I can put in there
17:25:57preglowamiconn: i fully expect it to be, there was nothing in the changelogs for 2.18, even, to indicate it's been fixed
17:26:01 Join J3TC- [0] (
17:26:04preglowamiconn: why did you check 2.17, btw?
17:28:06 Join Arathis [0] (
17:28:25pondlifeNico_P: The H300 has quite a large screen compared to some
17:28:56Nico_Pwhat screen should I take as reference? the nano's?
17:29:09*Nico_P looks for screen resolutions
17:29:34LloreanNico_P: The nano's screen has more lines than the h300
17:29:53 Quit amiconn (Nick collision from services.)
17:30:00 Join amiconn [0] (n=jens@rockbox/developer/amiconn)
17:30:06Nico_PLlorean: seriously? the H300 is 220x176
17:30:14LloreanErr, sorry, thought I was typing h100
17:30:18LloreanNano is 176x132
17:31:05Llorean*most* of our bitmap screens (all but Archoses and the c200 I believe) are 128 pixels or more tall
17:31:08Nico_Pthe c200 has the least lines I think... 80 px
17:31:20LloreanOh, and iPod mini
17:31:26Nico_PLlorean: I'm only interested in SWCODEC targets for the moment
17:31:36Nico_PI'll try on the c200
17:32:10 Join Crash91 [0] (n=evil91@
17:32:50Nico_P80 pixels is going to be tough... is it a problem if I go over that?
17:33:36Nico_Pmaybe I could add more information in ifdefs
17:33:41n1spreglow: when doing the shifts with mac is positive values shifting left and negative right? also are these shifts free and how many can I do at once? :-)
17:33:45*preglow wonders why amiconn always has nick collisions...
17:33:57preglown1s: afaik, yes, yes, and only one
17:34:33preglowit's pretty much intended exactly for what i did, adjusting the numerical range up or down to get free clipping, so it's pretty limited
17:34:42n1spreglow: but it shifts before adding to the acumulator, right?
17:34:47preglown1s: yep
17:35:03n1shmmm, might be useful
17:35:26preglown1s: it's important to remember the accumulator is 40 bits, so shifting does never overflow anything
17:35:42preglowanswer can be max 33 bits after shift
17:36:04preglown1s: also, it's important to remember it only work in integer mode, the shift does nothing in frac mode
17:36:15 Join WalkGood [0] (i=WalkGood@unaffiliated/walkgood)
17:37:12 Join lazka [0] (
17:37:18n1saha, so it won't work for tremor then
17:37:50preglowthe free shift would be quite useful in both tremor and musepack
17:37:58preglowsince they expect the answer unshifted
17:38:11preglowand mac does an implicit shift in frac mode
17:39:16 Join criznach [0] (
17:39:43 Quit Crash91 ("Bye Bye!")
17:40:18 Quit JdGordon (Remote closed the connection)
17:40:20 Join bluey^ [0] (
17:40:53pondlifeDo any voice users find the "Loading" splash annoying - i.e. the way it says "Loading" whenever you select a new track?
17:44:19 Quit FOAD ("I'll be back")
17:47:44 Join FOAD [0] (
17:49:45 Quit Arathis ("Bye, bye")
17:50:07 Join Arathis [0] (
17:50:40preglowi would...
17:50:58preglowpondlife: as in every time you play a track from playlist or file viewer?
17:51:32 Quit bluey- (Read error: 113 (No route to host))
17:51:36preglowpondlife: that'd annoy me no end, perhaps it could say "loading" if no song was playing from before. if music is currently playing before choosing, the only audible feedback you should need is the current music stopping immediately
17:51:46preglowwhich it also currently doesn't do, and someone well-versed in playback.c should fix that
17:52:13pondlifeIt's something that's annoyed me for a while. I'd rather just have a keyclick and no splash ever..
17:52:31preglowsounds fine by me
17:52:45preglowhowever, some feedback is good for blind users, and i think the scheme i mention might work
17:53:03pondlifeI'm in MoB-mode at the moment, not on SVN code
17:53:07preglowthe thing about current music stopping at once when you select a new trick needs fixing...
17:53:32preglowthe music just going on until the new track is loaded feels sluggish
17:55:48 Join freqmod__nx [0] (
17:56:14 Quit freqmod_nx (Read error: 104 (Connection reset by peer))
18:02:22 Quit kugel (Read error: 110 (Connection timed out))
18:11:01 Quit barrywardell (Remote closed the connection)
18:12:22preglowwell, if crossfade is disabled...
18:12:24preglowi never use the thing
18:12:29 Join Ebert [0] (
18:25:25 Quit Ebert ()
18:25:51 Quit Soap (Read error: 104 (Connection reset by peer))
18:25:57 Join Soap_ [0] (n=Soap@rockbox/staff/soap)
18:26:22 Join eXpL0iT [0] (n=exploit@
18:28:09eXpL0iTDo you guys have any idea where I can find some old iPod 5.5G firmware ?
18:28:29eXpL0iTespecially looking for iPod_25.1.2
18:28:31markunnot me
18:28:50markungoogle didn't find anything?
18:29:10eXpL0iTno the only thing i can find is some iPod_25.1.2.1
18:29:23eXpL0iTbut thats the version i've
18:30:10eXpL0iTi need to downgrade it :/ normally iTune saves the old firmware before flashing but i formatted my pc lately so lost my old firmware backup
18:31:43eXpL0iTso if you've an ipod video 30gb (pirates of the carabean) plz check C:\Documents and Settings\YOURUSTERNAME\Application Data\Apple Computer\iTunes\iPod Software Updates
18:31:51eXpL0iTthx in advance :) !!
18:33:00 Quit Soap_ (Read error: 104 (Connection reset by peer))
18:33:08 Join cooz [0] (
18:33:20 Join Soap [0] (n=Soap@rockbox/staff/soap)
18:33:40scorche|weXpL0iT: what does that have to do with rockbox?
18:35:27 Quit TMM ("Ex-Chat")
18:38:37 Quit kubiix ("Miranda IM! Smaller, Faster, Easier.")
18:39:14 Join kubiix [0] (
18:39:16 Join miepchen^schlaf [0] (
18:39:26markunscorche|w: that they are both firmwares :P
18:41:29 Nick fxb is now known as fxb__ (
18:43:35pondlifeNico_P: Just found a nasty MoB bug
18:44:00pondlifeIf you play a track that is bigger than the buffer, it stalls when it runs out of data
18:44:02Nico_Ppondlife: :( what is it?
18:44:10Nico_PI'll check
18:44:28pondlifeI'm trying a 35MB MP3 with a 26MB buffer
18:44:50Nico_PIt's probably just a matter of tweaking the buffering loop
18:45:00pondlifeOK, gotta go
18:45:01 Part pondlife ("Gone")
18:48:40 Quit bluey^ ("Leaving")
18:53:09coozhi, is there place where i can check progress of implementing sansa rockbox native usb handling?
18:54:13markuncooz: just the IRC logs right now
18:54:50coozmarkun: any hints where to search it?
18:55:23krazykitsearch for "zagor"
18:56:15coozis it a guy who's working on it?
18:57:14markunor usb
18:58:45 Quit parafin ("So long and thanks for all the fish")
18:58:51 Join parafin [0] (
19:00:34 Join ilgufo [0] (
19:03:38 Join FOAD_ [0] (
19:03:47 Quit FOAD_ (Client Quit)
19:11:34 Join pondlife [0] (n=Steve@rockbox/developer/pondlife)
19:11:42 Quit pondlife ("Read error: 110 (Connection slimed out)")
19:12:39 Quit J3TC- (Read error: 110 (Connection timed out))
19:14:47 Join pondlife [0] (n=Steve@rockbox/developer/pondlife)
19:15:04 Join mirak [0] (
19:15:58Nico_Ppondlife: I just tried a long track without a problem
19:16:14pondlifeHmm, it's reproducible here on H300
19:16:24pondlifeI did do some fast forwarding
19:16:27***Saving seen data "./dancer.seen"
19:16:56Nico_Pdoes the "useful" value behave as expected?
19:18:46Nico_Poh I think I know, although it's strange I didn't have the same thing... maybe add a call to fill_buffer() just after call_buffer_low_callbacks() when num_handles > 0 && useful_data() < conf_watermark
19:21:29 Quit ilgufo (Client Quit)
19:23:38Nico_Phmm even that shouldn't be necessary
19:23:46 Join Lear [0] (i=chatzill@rockbox/developer/lear)
19:25:06LearHm, playlist catalog is a bit ... odd. Never strips extension, ignores "ask bookmark" and "warn about modified playlist" settings.
19:27:01LloreanI thought once a playlist was loaded from the catalog, it's just a playlist...
19:27:27LearWhat do you mean?
19:27:53LloreanI dunno, really
19:28:06LloreanI'm uncertain why "playlist catalog" would interact with "ask bookmark" or "warn about modified"
19:29:07LloreanI mean, for modified, isn't that only dynamic playlists anyway?
19:30:34 Join ompaul [0] (n=ompaul@gnewsense/friend/ompaul)
19:30:51 Quit petur ("work->home")
19:31:02 Join MethoS- [0] (
19:31:46LearWhen you use the catalog to start playback, behavior is different compared to doing the same thing from the file browser. Hence "odd".
19:31:59preglowamiconn: here?
19:34:06 Join hcs [0] (n=agashlin@rockbox/contributor/hcs)
19:34:06LloreanLear: Ah, I see what you mean.
19:41:12LearBtw, does the sansa OF insist on hiding the playlists catalog too?
19:42:02 Quit pondlife (Read error: 110 (Connection timed out))
19:43:01amiconnpreglow: Here now
19:43:51amiconnpreglow: I have configured my client so that if I drop from irc (regular disconnect one per day by the provider, wlan hiccup etc), it first tries to ghost the nick before joining the channel
19:44:13amiconnThat way I avoid the _ ___ nicknames
19:44:16preglowahh, k
19:44:40preglowamiconn: but yeah, did you check 2.18 of binutils?
19:44:51preglowamiconn: also, where is the bug in the source?
19:45:03amiconnNo, I'm checking a 2.16.1 with my own patch
19:47:44 Join bertrik [0] (
19:48:14amiconnBuilding binutils took quite some time, but now I am testing with msac.w in filters_cf.S
19:48:52preglowwouldn't it be worth checking if the newest binutils has fixed the bug first?
19:48:54 Join pixelma [0] (i=pixelma@rockbox/staff/pixelma)
19:49:26amiconnHmm, what I found was only one part of the bug...
19:49:43preglowthat was my problem with the gas source, it's spread all over the place
19:49:49amiconnThe 2 factor registers are now correct
19:50:00amiconnThe parallel load address is not
19:50:45amiconnThe factor register bug is in m68k-opc.c, lines 1652..1654 (2.16.1)
19:51:23 Join Zagor [0] (
19:52:47amiconnAh, and the other one too (another letter in that string), I just overlooked it before
19:55:35 Join ilgufo [0] (
19:55:48 Join BigBambi [0] (n=alex@rockbox/staff/BigBambi)
19:56:33 Quit eXpL0iT (Remote closed the connection)
19:56:48amiconnpreglow: Evencvs binutils still have that bug
19:57:02amiconn*Even cvs
19:57:13preglowamiconn: right, are the other mac variants fine?
19:57:30preglowthen it's time to send the binutils dudes a patch
19:57:37amiconnThe bug only hits msac.w with parallel load
19:57:42preglowlike i thought
19:57:47preglowi've pretty much used all mac variants now
19:57:52amiconnmsac.l with parallel load is okay, as are all other variants
19:58:30amiconnMy patch is working now
19:58:38preglowbut the question remains, what should we do in the meanwhile? push a patch to all our build server until we start using a newer bintuils?
19:59:36amiconnI'm not even sure whether we could use 2.17 or higher in conjunction with gcc 3.4.6
19:59:45preglowno, it would have to be after 4.3
19:59:59preglowit's still some time off
20:01:36preglowbut seeing as how we're currently just hit in one spot, we could just ignore this entire issue until we start using a newer binutils (which will hopefully have the feature fixed)
20:01:46 Join technobeast [0] (n=Georgiev@
20:02:53amiconnWe could also require our patch, as we require some gcc patches
20:03:12technobeastHi, there... I have an iPod video 30 GB... and I think I didn't do something as it had to be.. and now it won't boot. I just hear this horrible clicking sound when it tries to boot...
20:03:18technobeastIs there anyone that can help me
20:03:22 Join Bagder_ [0] (
20:04:41preglowamiconn: i know, but i just don't know if the hassle of updating all the build servers is worth it
20:04:42Lloreantechnobeast: There shouldn't be a clicking noise even if you do do something wrong. What all have you done so far, and have you tried simply restarting the device and putting it into disk mode?
20:05:06preglowamiconn: it'll take several days, all to shave off one instruction in the speex codec
20:05:28preglowi say it's not worth it for now, i can always hard code the instruction too, .long style
20:06:08pregloweight and ten .longs in the middle of the code...
20:06:14technobeastI just succeeded to put it in disk mode.. as I'm talking to you
20:06:24technobeastOtherwise the battery was very low
20:06:32technobeastAnd I think this is the main problem
20:06:36technobeastIt just wouldn't boot
20:06:53technobeastI saw the apple logo.... and then it started clicking as it tries to power on
20:06:57LloreanYes, a very low battery will prevent the disk from being able to spin, and can cause clicking noises when it tries...
20:07:14technobeastAfter 5-6 clicks it writes "Please wait. Very low battery"
20:07:19technobeastAnd this goes on forever
20:07:33preglowamiconn: no wonder i didn't find the bug....
20:07:45amiconnWas quite easy to find in fact
20:07:58 Part flynux
20:08:10amiconnYou just need to put the mac.w and msac.w parts side by side and look close enough
20:08:23*amiconn likes Notepad++ for having dual view :)
20:08:25 Join SMille [0] (i=53e279b4@gateway/web/cgi-irc/
20:08:31*preglow hugs gvim
20:09:06amiconnAnd the file can be found by greping for 'msac'
20:09:09preglowamiconn: anywho, binutils has a bugzille :-( i guess it should go there
20:09:32preglowdiffed against cvs, of course
20:09:40amiconnI will put it there only if they're not like the gcc people...
20:09:47preglowwell, we don't know that
20:09:55preglowthis isn't even a bug report, this is a fix
20:10:00preglowthey can't _not_ commit that
20:10:32SMilleyo! AM trying to compile a build of rockbox with the espeak pluggin but i cant get it to worke :/ The thing is that the source file dont have an file extention :/ some one know what i chell do?
20:10:50preglowsource file? what source file?
20:10:53 Part gryzor
20:11:27technobeastSorry for the bothering, guys....
20:11:41technobeastI think I made it to restart after a while in "disk mode"
20:12:12 Part technobeast
20:12:22preglowthose zips don't even work here
20:12:30preglowthey just contain a ton of empty subdirs
20:13:49SMilleok :/ maybe that is the problem, i can ectract em using winrar, but it my be curupet files thats why no extention :/
20:14:09SMilleDarn i really whould like to install a TTS
20:14:15 Quit MethoS- (Connection timed out)
20:14:26preglowamiconn: so ldmia/stmia doesn't help much on arm9?
20:15:17 Join petur [0] (n=petur@rockbox/developer/petur)
20:15:25amiconnNo, as they need n cycles for n registers
20:15:51amiconnAnd ldr, ldrh and ldrb are single cycle - unless you use the loaded value in the next instruction
20:16:01preglowyeah, fancier pipeline
20:16:22amiconnSo it does make sense to use ldmia if you're going to process the values immediately
20:16:30amiconnAnd of course it makes more compact code
20:17:31preglowhow many cycles before i can use the fetched value? two?
20:17:54amiconnSo using it immediately produces a one cycle stall
20:18:11amiconnIirc with ldmia it's 2 cycles, but from when the register was loaded
20:18:28amiconnSo with ldmia it makes sense to process the first loaded registers first
20:19:14amiconnIn fact I wonder why arm7tdmi doesn't have that kind of pipelining - the much older SH1 does
20:19:33 Join hannesd [0] (
20:20:12 Quit Bagder (Read error: 110 (Connection timed out))
20:21:15preglowthey probably did what required the least transistors
20:21:23preglowarm cores are very minialistic
20:21:25 Quit hcs (Read error: 113 (No route to host))
20:21:37amiconnThe bug is not in their tracker as well....
20:22:08amiconnMrf, I need to create an account :(
20:23:42 Join eXpL0iT [0] (n=exploit@
20:24:09 Join TotallyInfected [0] (n=ebola@
20:24:25SMillesome one in here have the espeak plugin?
20:25:30preglowamiconn: even we require that nowadays...
20:25:49amiconnYeah I know...
20:28:52*rasher whispers OpenID
20:33:12 Join MethoS- [0] (
20:38:08preglowamiconn: does objdump reuse the gas tables?
20:38:25amiconnI don't know
20:39:10amiconnDid you test speex with a patched gas?
20:39:29amiconnI compiled it, but didn't test it, I just disassembled the .o
20:39:54*amiconn needs to do some commits
20:40:22lostlogicanyone have sid files want to make sure I didn't break anything in a small patch?
20:40:42preglowamiconn: could you please just upload the resulting speex.codec some place so i don't have to recompile binutils?
20:40:53lostlogicscratch taht, broke flac, will try later.
20:40:56amiconnI've built for x5
20:41:03amiconnYou need h1x0?
20:41:05preglowh120 would be cool
20:41:57n1slostlogic: the HVSC contains loads of sidfiles ;)
20:43:04preglowno shit
20:43:37CalcipherHave any of you been able to generate voice files in debian using sapi5? well the rb debian on vmware that is
20:44:23lostlogicn1s: HVSC?
20:44:48Lloreanlostlogic: High Voltage Sid Collection
20:45:04LloreanSome ridiculously large number of sid files all in one nice archive for download.
20:45:22LloreanEasiest way to very quickly have a collection of music that will break the Rockbox database
20:46:49rasherCalcipher: sapi is a windows thing
20:47:10 Join japc [0] (
20:48:38Calcipherwell IIRC someone mentioned being able to generate voice files under cygwin
20:49:22LloreanCygwin is basically like a replacement command prompt. It's still more or less windows, it just lets you carry out commands in a linux style
20:50:01Calcipherwell, then I atleast would like to have access to my NTFS volumes under the vmware session, is that possible?
20:50:37 Quit Zagor ("Client exiting")
20:50:49 Part Llorean
20:50:59Calcipherthanks Llorean, I was under the impression it was similar to the vmware method
20:51:47 Join Zagor [0] (
20:52:07 Join ze_ [0] (
20:52:23 Quit ze (Read error: 104 (Connection reset by peer))
20:52:27 Nick ze_ is now known as ze (
20:53:08CalcipherIf I can access my NTFS volumes I can test a different tts system and try to generate some voice files
20:54:35LearAnyone familiar with "changing_dir" in playback.c? :) It's ugly, and I can't see the point in having it...
20:55:48preglowamiconn: sounds 100% good
20:58:45 Quit cooz ()
20:58:57lostlogicsweet, I've got sids.
21:00:58 Join webguest29 [0] (i=525163c1@gateway/web/cgi-irc/
21:01:53preglowwebguest29: ok
21:02:00webguest29hmmm i have a problem with some of the e200 themes.... some of them have bug while showing the time and date can someone help?
21:02:08*Nico_P is itching to commit MoB
21:02:56preglowNico_P: how's it working these days?
21:03:04n1sNico_P: everything working A-OK ?
21:03:05Nico_Ppreglow: well!
21:03:30Nico_Pn1s: I've fixed pretty much all known bugs
21:03:41Nico_Pin MoB that is, not those from SVN
21:03:41n1sNico_P: so should I test more?
21:04:27 Quit jgarvey (Read error: 110 (Connection timed out))
21:04:28Nico_Pn1s: well yeah you can tell me if you get rebuffering problems with long tracks... pondlife reported that but I don't have any problems
21:05:04 Quit webguest29 (Client Quit)
21:06:59 Join jgarvey [0] (
21:07:06n1sNico_P: i don't think i have any tracks that long that aren't flac but those already had the rebuffering problem in svn :-/
21:07:21Nico_Pwhat problem?
21:07:51n1sthat they pause when rebuffering because the audio buffer empties
21:08:07n1stoo fast, ie the watermark is too low
21:08:20Nico_PI'd be curious to know if this still happends with MoB
21:08:33n1sNico_P: ok, I'll test
21:09:28 Quit miepchen^schlaf ("Verlassend")
21:11:59 Part WalkGood
21:12:15*lostlogic starts abusing MoB with his hackified build too
21:13:03 Join Sedgewick [0] (
21:14:28 Join miepchen^schlaf [0] (
21:14:33 Join meoblast001 [0] (
21:16:13 Join moos [0] (
21:16:31***Saving seen data "./dancer.seen"
21:17:38amiconnCurrent rockbox crashes immediately after dircache scan on 2nd gen :(
21:17:56meoblast001does anyone know why only on the WPS i get the cool effects for my battery bar and stuff?
21:18:16meoblast001on menu's its just shaded black
21:19:08 Quit enakiejlo (Read error: 104 (Connection reset by peer))
21:19:23lostlogicmeoblast001: it's only configured by the .wps file so it is only customizable on the wps.
21:19:29 Quit FOAD ("I'll be back")
21:19:52 Join FOAD [0] (
21:20:06meoblast001lostlogic: is there a reason for that, if theres gonna be effects somewhere there should be effects everywhere
21:20:19 Join anakiokoi [0] (i=0@
21:20:21lostlogicmeoblast001: because only the wps is customizeable in rockbox at this time...
21:20:29amiconn..and with dircache disabled, it crashes immediately after boot
21:20:37lostlogicNico_P: I was able to data abort by starting a playlist buffering and immediately skipping.
21:20:43*amiconn retries with an official build
21:20:48meoblast001lostlogic: do they have plans on doing anymore?
21:21:07lostlogicmeoblast001: not that I know of
21:21:20amiconnHmm, official build works...
21:21:26*amiconn doing full rebuild
21:21:32lostlogicNico_P: I'll try without my hacky-hacks and see if it's happier.
21:21:49meoblast001i sent a feature request but it wasnt the clearest, since i was new to rockbox when i sent it (about 2 weeks ago)
21:22:08preglowamiconn: reported the bug yet?
21:22:14lostlogicmeoblast001: yes, I remember when you asked then.
21:22:29meoblast001i was told that there is a patch to do that
21:22:32meoblast001is that true?
21:23:12amiconnpreglow: Btw, it looks like you can also use << and >> for the mac instructions with shift
21:23:30amiconn(just literal << and >>, without numbers)
21:23:37lostlogicmeoblast001: beats me −− have you looked on the patch tracker?
21:23:52meoblast001yeah, but this would be my first time using one
21:24:12 Join Frazz [0] (
21:24:14lostlogicwell first then you need to learn how to build your own rockbox.
21:25:09preglowamiconn: cool
21:25:15preglowamiconn: i'll try
21:26:26preglowamiconn: works peachy, i think i prefer that too
21:26:42*amiconn apologises for the split commit
21:28:32 Join Llorean [0] (
21:28:49n1sNico_P: did you remove the buffering debug screen?
21:29:20*n1s listens to lamb of god in flac to test now :-)
21:29:48 Join kugel [0] (i=kugel@unaffiliated/kugel)
21:29:49lostlogicNico_P: so far can't reproduce the problem I mentioned so it was specific to the way I built.
21:29:56Nico_Pn1s: I replaced the audio debug screen with the buffering one
21:30:02Nico_Plostlogic: ok
21:30:58meoblast001lostlogic: if my feature is labeled Normal, does that mean it's been looked at?
21:31:15n1sNico_P: I'm just getting paranoid I guess but is it called view audio thread still and looks exactly like before?
21:31:37Nico_Pn1s: have you taken the latest snapshot?
21:32:00n1sNico_P: I think so, last commit 20 hours ago
21:32:01Nico_PI changed it a few commits back
21:32:18Nico_Pthat's too long ago
21:32:51n1si got the latest from here is that the wrong one?
21:33:23 Quit MethoS- (Read error: 101 (Network is unreachable))
21:33:41amiconneh? My own builds throw data aborts all the time while official ones work???
21:35:38*amiconn reconfigures and rebuilds once more :/
21:37:03 Quit RaZorbacK ("MegaIRC v3.90")
21:37:08n1sdoh! my snapshot doesn't have buffering.c so i guess it's just some coupy of svn
21:37:31n1sNico_P: could you give me the correct link that site is impossible to navigate
21:38:17Nico_Pclick snapshot on the latest commit
21:39:02n1sbtw I could reproduce the bug with the other build so it's still in svn
21:39:22 Join haemmy [0] (n=stefan@
21:40:19amiconnDoes the m:robe 500 have an lcd remote??
21:41:11meoblast001if my feature request is labeled Normal, does that mean it's been looked at?
21:41:24n1smeoblast001: no
21:41:38meoblast001nls: so what does that mean?
21:42:28moosNico_P: hi, planned to commit your monster MoB patch this night?
21:42:37 Join troxor [0] (n=boyo@unaffiliated/troxor)
21:43:03Nico_Pmoos: it should soon be committed... but "monster"? :p
21:43:30mooshehe, monster patch, cause lot of changes ;)
21:43:36n1smeoblast001: priority doesn't mean anything for feature requests
21:43:45Nico_Pmoos: tested it?
21:44:07meoblast001nls: so how does the priority get determined if no one looked at it
21:44:23n1smeoblast001: the feature requests are just for inspiration and ideas for devs nothing says your particular idea will ever be implemented
21:44:26moosNico_P:not found the oportunity yet. But when it will be svn, we will all test for you :)
21:44:42Lloreanmeoblast001: Someone may have looked at it for the purpose of determining the priority, but that could be someone who has absolutely no interest in the idea and is just keeping the tracker organized.
21:45:17meoblast001so what tells me whether the development team has looked at it?
21:45:32LloreanFirstly, there really is no "development team"
21:46:00meoblast001ok..... ppl who help build the code then
21:46:03moosNico_P: are there any remaining bugs?
21:46:04LloreanThere's a list of people who can make direct changes to Rockbox, and then there's everyone else who can submit their patches for approval.
21:46:05amiconngah! This is getting annoying
21:46:19amiconnI can't make a working build for 1st/2nd gen anymore :(
21:46:43Lloreanmeoblast001: Everyone who's interested can help build the code. That's the whole point. There's a list of people who decide what goes in or not, but feature requests are often filled by random people who are interested in helping out, and probably are more often because the people who have direct access are usually working on bugs or their own ideas.
21:47:06Nico_Pmoos: not that I know of
21:47:34moosNico_P: what is stoping you then? screaming? ;)
21:49:20rasheramiconn: pretty sure the mrobe has an lcd remote
21:49:49*amiconn craps working dir and does a fresh checkout :[
21:50:59n1samiconn: make veryclean, kill ccache?
21:51:09amiconnNo ccache, this is cygwin
21:52:09 Join RaZorbacK [0] (
21:52:20 Nick parafin is now known as parafin|away (
21:53:01 Quit haemmy ()
21:53:02bertrikCygwin can also use ccache AFAIK
21:53:26Calcipherbertrik, how did you update your sansa bootloader
21:54:03CalcipherI'm asking because I'm using an R series, and the original install varies from the original e200 install
21:54:16bertrikI built a bootloader, copied the .mi4 over to where my sansapatcher lives and installed it with sansapatcher -a
21:54:42bertrikno idea about the R series
21:54:59amiconnbertrik: Yes, but it's actually slower than not using it
21:55:00Calcipherhm, ok thanks
21:55:12n1sNico_P: my flac playlist has rebuffered without pauses twice now :-) however I can't tell how close we are to buffer underrun because the screen seems to stop updating when the diskreading starts but I estimate something like 1-2 seconds of compressed audio left...
21:55:22 Quit RaZorbacK (Read error: 104 (Connection reset by peer))
21:55:38 Join RaZorbacK [0] (
21:55:40amiconnn1s: The low watermark for flac is probably too low
21:55:59LloreanIt's been too low in the past, too, hasn't it?
21:56:02 Quit RaZorbacK (Read error: 104 (Connection reset by peer))
21:56:14 Join RaZorbacK [0] (
21:56:39n1samiconn: yes, in svn it actually underruns sometimes
21:57:11n1sLlorean: yup I even put that bug in the tracker last year :-)
21:57:23amiconnI don't expect this to be part of MoB; and the fact that MoB doesn't underrun says something about it...
21:57:33CalcipherI tried using rbutil to install the bootloader, it said it couldn't find any sansas, yet it was able to update my install of rb
21:57:34amiconnBut the watermark still needs to be raised then
21:57:55Lloreanamiconn: Well, if MoB buffered just a little bit faster then it wouldn't underrun too, right?
21:58:00n1sNico_P: can't get it to lock up by skipping like a madman anymore either
21:58:03amiconnLlorean: Yup
21:58:08LloreanEven if it "says something" it might not be a *bad* something ;)
21:58:23amiconnBut 1..2 seconds is too low to reliably avoid underruns
21:58:33amiconnI didn't say that it's bad...
21:58:48n1sbut when skipping a lot of tracks it takes several seconds for the wps to update so it is very difficult to know how far you went in the playlist
21:58:58LloreanSorry, the phrase "it says something about Noun" usually has negative connotations around here at least. Misinterpretation by me.
21:59:09amiconnDid someone compare buffering times with svn vs. MoB and a demanding codec?
21:59:24LloreanIs there any way to determine the maximum bitrate of a FLAC file?
21:59:39LloreanOr are you stuck with the average?
22:01:22 Quit RaZorbacK ("MegaIRC v3.90")
22:01:37 Join RaZorbacK [0] (
22:01:52Nico_Pn1s: good news :)
22:02:04Nico_Pamiconn: AFAIK there has been no comparisons
22:02:48Nico_Pamiconn: btw, binsize increase is only a bit more than 300 bytes, most of it being additions to the debug screen
22:03:25amiconnMeh, something is definitely hosed here :(
22:04:38 Join Ebert [0] (
22:06:07n1sNico_P: when starting to play an nsf file in a dir with a couple of hundred others it takes several seconds before it starts playing and seems to try to buffer too much before beginning
22:06:58Nico_Pn1s: if you only add the single file to the playlist it starts immediately?
22:07:39n1sit's much faster at leas
22:08:00Nico_Phow did SVN use to behave on this? faster?
22:08:23n1sNico_P: don't remember actually, will test
22:08:26 Join webguest69 [0] (i=525163c1@gateway/web/cgi-irc/
22:09:07webguest69How i can make GBA Emulator for mp4?
22:09:15webguest69or where i can download it?
22:09:22n1sNico_P: yup svn is a lot faster
22:09:38Lloreanwebguest69: This is #rockbox, and is for discussion of Rockbox.
22:10:08webguest69it's plugin..... Llorean
22:10:26webguest69i mean the plugin for sansa e200 can only run GB and GBC
22:10:33webguest69but i want to run GBA roms
22:10:37webguest69how i can do it?
22:10:39LloreanYes, the Rockboy plugin is only a gameboy and gameboy color emulator
22:10:47LloreanNobody has ported a GBA emulator, and the Sansa is too slow anyway
22:11:51webguest69ohh ok
22:12:08webguest69btw why the sound on some games is bugged
22:12:20webguest69i mean gbc or gb games
22:12:34LloreanBecause it still needs work.
22:12:50 Join J3TC- [0] (
22:13:11webguest69and ?0_0
22:13:25LloreanAnd what?
22:13:46bertrikI guess it stays bugged until someone fixes it
22:13:46webguest69what u meant when u said still need to work?
22:13:59webguest69ohhh ok thanks
22:14:13 Quit webguest69 ("CGI:IRC")
22:16:32 Nick Bagder_ is now known as Bagder (
22:16:58 Quit Isolinear (Read error: 110 (Connection timed out))
22:17:01 Join bluebrother [0] (i=aORfJPlW@rockbox/staff/bluebrother)
22:19:46 Quit amiconn (Nick collision from services.)
22:19:53 Join amiconn [0] (n=jens@rockbox/developer/amiconn)
22:20:35 Join cooz [0] (
22:23:41 Quit Toxicity999 (Read error: 104 (Connection reset by peer))
22:24:15 Quit Frazz ("Leaving")
22:24:45 Join Toxicity999 [0] (n=bryan@unaffiliated/Toxicity999)
22:25:57 Quit ender` (" Some people like my advice so much that they frame it upon the wall instead of using it. -- Gordon R. Dickson")
22:26:18 Join ender` [0] (
22:29:57 Quit Lear ("ChatZilla [Firefox]")
22:36:41 Quit Toxicity999 (Read error: 104 (Connection reset by peer))
22:37:37 Join Toxicity999 [0] (n=bryan@unaffiliated/Toxicity999)
22:38:29 Quit moos ("reboot")
22:39:13 Join donutman25 [0] (n=chatzill@
22:39:28amiconnEek. The range decoder in libdemac does (at least) 2 divisions per sample (iiuc).
22:42:28 Join moos [0] (
22:44:01Zagorany Gigabeat S owner here?
22:44:05 Join nomel [0] (i=4782332a@gateway/web/cgi-irc/
22:44:08LloreanZagor: I own one.
22:44:14LloreanBut I haven't tried anything Rockbox related with it yet.
22:44:18nomeli'm looking at the fm debug screen
22:44:25 Quit nicktastic ("Leaving")
22:44:30ZagorLlorean: aha, so you can't run some test code for me?
22:44:35nomelthe mss_fm (radio frequency i'm assuming) is not matching what the fm screen is saying.
22:44:40nomelthe debug mode shows the correct screen.
22:44:46nomelerr...correct frequency.
22:44:54LloreanZagor: I don't even know how the others run unsigned code, yet. :)
22:45:05LloreanZagor: You may want to try #gigabeat, they seem to do more of their talking in there.
22:45:18Zagoroh, ok
22:45:28Nico_PLlorean: they do run code
22:45:41LloreanNico_P: I know, I said *I* don't know how they do. :)
22:46:01Nico_Poh, I misunderstood. it involves patching a firmware update
22:46:08LloreanIf I recall they even had it writing to LCD before aliask messed his up
22:47:00 Quit TotallyInfected (Remote closed the connection)
22:47:34kkurbjun_amiconn: teh mrobe:500 does have an lcd remote 79x16
22:47:38 Join TotallyInfected [0] (
22:48:04amiconn*really* tiny then.
22:48:27nomeland incrementing the frequency in 0.05 MHz steps causes the actual frequency to go way off...for instance 94.7, 76.0, 89.2, 82.5 instead of 98.5, 98.55, 98.6, etc.
22:49:24kkurbjun_yes, not much of a display, just 2 lines of text really.. it is a bitmap lcd driver though
22:50:11 Quit SMille ("CGI:IRC")
22:50:43 Join Shaid` [0] (
22:50:55 Join SMille [0] (i=53e279b4@gateway/web/cgi-irc/
22:51:28nomelthis is for the c200
22:51:33 Join haemmy [0] (i=stefan@
22:54:00 Quit SMille (Client Quit)
22:54:04 Quit bertrik ("bye")
22:56:28kkurbjun_amiconn: are you thinking about getting an m:robe 500? :)
22:57:20amiconnI just wondered about the remote bitmap format definitions in configure
22:57:53amiconnAnd m:robe 500 is touchscreen, so a big no-no for me
22:58:35LloreanThe m:robe has a camera
22:58:51kkurbjun_oh, I gotcha, you don't /have/ to use the touch screen with the remote though. Why don't you like the idea of a touchscreen player?
22:59:05 Quit petur ("switch")
22:59:05kkurbjun_yeah, it has a camera builtin too
22:59:08 Join p3tur [0] (n=petur@rockbox/developer/petur)
22:59:23 Quit TotallyInfected (Remote closed the connection)
22:59:37kkurbjun_I don't think we'll be supporting that any time soon :)
22:59:42amiconnI like to be able to operate the thing without looking at it
23:00:03 Join TotallyInfected [0] (n=ebola@
23:00:16amiconnThe touchy stuff on some other devices (ipods, H10) are already enough annoyance
23:00:17kkurbjun_we still have to figure out how to get the DSP part to forward sound on the i2s interface..
23:00:33 Quit ilgufo ("So Long, and Thanks For All the Fish -")
23:01:02kkurbjun_I can understand that.. the touch screen is really fast to navigate if you are looking at it with the limited support we have for it now.
23:01:10 Nick cooz is now known as cooz` (
23:01:49kkurbjun_especially with a 640x480 screen, you can see so many items at once and you don't have to scroll through all of them to pick one
23:02:08*amiconn prefers small screens
23:02:20 Join webguest32 [0] (i=525163c1@gateway/web/cgi-irc/
23:02:32amiconnAnd a transflective screen too. Most colour screens are not.
23:03:16kkurbjun_hmm, well I can't really argue with your preferences can I :)?
23:03:17webguest32I have a question, is it possible playing GB or GBC while music is playing simulteniously?
23:03:39webguest32i mean playing GB GBC games....(plug ins )
23:03:40kkurbjun_the m:robe would be a terrible player given those requirements
23:04:23webguest32please help
23:04:31webguest32e200 device
23:04:46kkurbjun_the transflective screens are the ones that show in the sun well?
23:04:56kkurbjun_webguest32: only on the gigabeat
23:05:20kkurbjun_none of the other players support playing music with gameboy games
23:05:48amiconnAnd just 20GB is a bit small for a HDD based player today...
23:05:52*amiconn would expect 100GB+
23:06:34 Quit Ebert ()
23:06:35amiconnkkurbjun_: Yes, transflective screens are readable with just ambient light. All monochrome/ greyscale LCDs are, but most colour LCDs are not
23:06:52amiconnThe only exceptions among the rockbox targets being the iPod Video and Nano
23:07:50n1sNico_P: from what I can see, MoB seems solid now :-)
23:07:57Nico_Pncie :)
23:08:00Nico_Por nice
23:08:10moosn1s: same feeling here
23:08:53 Quit webguest32 ("CGI:IRC (EOF)")
23:08:55kkurbjun_I enjoy the target if nothing else for the difference in the user interface from anything else we support right now.. JdGordon's done all of the work for the touchscreen so far with Cat's existing code.
23:09:28moosNico_P, n1s: just 20 minutes of use, but that seems stable enough for normal uses and svn imho
23:09:49kkurbjun_the DSP part is interesting as well - it's a shame the tools for it aren't free (or the docs)
23:09:50*Nico_P has been using MoB for everyday use for days now
23:10:14moosgo for commit then ;)
23:10:34LloreanBut make sure you have plenty of time to fix red on any targets you may not have tested on. ;)
23:10:48mooshehe :)
23:11:05Nico_PLlorean: indeed. I've tested a bit and I hop to limit the red
23:11:28moosNico_P: don't be chicken :P
23:11:37n1sNico_P: I vote for a commit too, there will always be bugs we missed but none too big it would think
23:11:41LloreanIt seems likely there won't be much, but things aren't as predictable as they should be.
23:11:57LloreanMuch red, that is.
23:11:59LloreanIf any
23:13:47 Quit haemmy (Read error: 104 (Connection reset by peer))
23:13:48 Join haemmy2 [0] (n=stefan@
23:13:59n1sa little red doesn't matter as long as you fix it
23:14:20moosthat's not like if we never have red ;)
23:14:45*Nico_P is preparing his svn tree
23:14:52Nico_Pnow what to say in the commit message
23:14:57moosand bugs are surely harder to fix
23:15:07mooswee wee
23:15:27lostlogicNico_P: I get data aborts when I try to run the codec thread on the COP, but that's beyond the scope of your stuff to fix :-P
23:15:51Nico_Plostlogic: indeed ;) still room for improvement then
23:15:54mooslostlogic: let's jhMikeS area a part
23:16:05lostlogicNico_P: commit away, it feels solid to me and performance seems tob e the same subjectively
23:16:14lostlogicmoos: yeah
23:16:17mooslostlogic: by the way, nice to see you back
23:16:28scorche|wdont say that!
23:16:31lostlogicmoos: I'm an illusion, don't fool yourself!
23:16:32***Saving seen data "./dancer.seen"
23:16:41mooshaha :D
23:16:44LloreanNico_P: Might want to mention in the commit message that there should be no user-visible differences, so people know they should be reporting things working odd?
23:16:59Nico_PLlorean: good idea
23:17:23lostlogicNico_P: the new combined buffering screen is beautiful... and costs 2% boost ratio in my test track compared tot he old :D
23:17:49lostlogicgood thing debug _performance_ is a critical measure :-D
23:17:57Nico_Pmaybe I should reduce it a bit.... I'm used to my gigabeat and sometimes forget such considerations
23:18:08lostlogicno no, it's fine
23:18:28lostlogicI was just joking around, who cares if the debug screen is expensive, it's debug
23:18:33 Quit haemmy2 (Client Quit)
23:18:45 Quit japc (Read error: 110 (Connection timed out))
23:18:53 Quit Shaid (Read error: 110 (Connection timed out))
23:18:53 Nick Shaid` is now known as Shaid (
23:19:14lostlogicsiht, just triggered that data abort!
23:19:20lostlogicI'm sure that's nto what you wanted to hear
23:19:28 Quit desowin ("use linux")
23:19:56Nico_Pindeed not :) what's the problem?
23:20:47 Join einhirn [0] (
23:21:53Nico_Plostlogic: should I stop writing the commit message?
23:22:42lostlogicNico_P: don't know what the problem is, but it happens when I skip backward and forward within a large playlist with tagcache and dircache off
23:22:53lostlogic(taht is I run a build w/o tagcache and w/o dircache)
23:23:16 Quit zicho (Remote closed the connection)
23:23:24 Quit J3TC- (Read error: 104 (Connection reset by peer))
23:23:46lostlogicwhether you commit any way is up to you :-P
23:24:02 Join J3TC- [0] (
23:24:04lostlogicI haven't seen the same problem when I run the with tc and dc 'normal
23:24:07Nico_PI think I'll still do it
23:26:46lostlogicand I'll keep hunting for my personal mysterious data abort
23:28:13lostlogicoh yeah, I forgot, I also turn off priority scheduling, wonder if that's the diff
23:28:22*Nico_P imagines the avalanche of bug reports tomorrow
23:28:53Nico_Plostlogic: I suggest you try disabling only one feature at once to see if any of them is responsible
23:28:59Nico_Pafter trying a regular build of course
23:29:01moosnice log message btw
23:29:29*jhMikeS gets here and sees "committed"...who or what? :)
23:29:42LloreanjhMikeS: MoB
23:29:44*Nico_P and MoB
23:30:03jhMikeS*G* ...and MoB too?
23:30:38*bluebrother wonders why MoB partially disables cuesheet support
23:30:55Nico_Pbluebrother: the part in audio_load_track is disabled but it will still work
23:30:57bluebrotherdoesn't that just seek to some file position?
23:31:16Bagder"It takes some actually effort on your part to hack it, but the documentation is truly world-class."
23:31:29Bagder(about rockbox that is =>
23:32:24 Quit amiconn (" sound borked again :(")
23:32:26bluebrotherdon't we write it "Rockbox"?
23:32:58*Nico_P hopes the red delta won't be too red
23:32:58LloreanYep, we do
23:33:00jhMikeSNico_P: yes...playback.c is like a bone that must be broken to be properly reset
23:33:53*lostlogic decides to try really really hard to data abort on a non-hacked build
23:33:54jhMikeSblow out the crossfade too while at it
23:34:02bluebrotherhmm ... how about finishing 3.0?
23:34:22Nico_Pwhoo, green! :D
23:34:26*Llorean cheers
23:34:49lostlogicNico_P: nice!
23:34:49mooslet's see for delta now ;)
23:34:50Bagdernice delta
23:34:58LloreanBagder: The delta's from the previous commit
23:35:02Nico_Pthere must be a problem
23:35:12LloreanOh, wait
23:35:16bluebrotherwow, impressive delta.
23:35:21Nico_Pthis can't be true
23:35:23lostlogicwow Nico_P how... what...?
23:35:35Bagderhover for the details
23:35:41Bagdermost gain is ram
23:35:49bluebrotherNico_P: you forgot to add the archos players to the delta ;-)
23:35:49Nico_Pah, that makes sense
23:36:02*n1s guesses on the removed static metadata struct
23:36:04Nico_Pit's metadata on buffer after all ;)
23:36:26*bluebrother wonders how big that struct was
23:36:31LloreanSee, now we just need deltas like that every day. :-P
23:36:35Nico_Pbluebrother: more than 1k
23:36:42Nico_Ptimes 32
23:36:44 Join barrywardell [0] (
23:36:49moosit's probably the new record :)
23:36:56bluebrotherthat's a lot
23:37:05 Join amiconn [0] (n=jens@rockbox/developer/amiconn)
23:37:12bluebrotherwe should really keep a graphical table of the size deltas ;-)
23:37:14moosNico_P: sure, nothing missing? :)
23:37:32Nico_Pmoos: I wondered at first, but the gain is in the RAM
23:37:36jhMikeSplayback.c is #if 0'ed out?
23:38:00n1sbut it really cut 1k of binary size on coldfire at least arm gained a bit
23:38:43Nico_Pn1s: really? 1k? \o/
23:39:19lostlogicok, I can data abort even when 'normal'
23:39:33n1slostlogic: killjoy! :P
23:39:57jhMikeSthe delta table is almost as wide as the target table
23:40:00lostlogiclaugh, only way to get me to participate in the project is to brake shit I use, so consider it a blessing :-P
23:40:15lostlogicbreak that is
23:41:27amiconnjhMikeS: Even wider here - wider than my screen...
23:41:34 Quit p3tur ("*plop*")
23:41:42*BigBambi scraps the MoB build he did 15 minutes qgo for q current build :)
23:42:06*bluebrother hands BigBambi an "a" :)
23:42:12BigBambireplace q with a where appropriate - dams switching between UK and FR keyboards
23:42:30Nico_PBigBambi: hehe, hard to get used to azerty? ;)
23:42:48BigBambiNico_P: My laptop is qwerty, and at work azerty
23:42:57BigBambiSo it is just when I swap I get lost
23:42:59Nico_Pouch, even harder
23:44:20*amiconn wants that ram saving on hwcodec :/
23:45:39bluebrotherinteresting, the "show track info" scren now shows ".bmark" as genre :o
23:46:10moosamiconn: at least, now that the playback code is less complex and more documented, maybe that will make an playback engine unification easier for someone
23:46:35Nico_Pbluebrother: from the WPSN
23:47:00lostlogicNico_P: my data abort appears to happen in find_handle
23:47:10bluebrotheryes, accessing the info through the wps context menu
23:47:45Nico_Plostlogic: ah, perhaps the recent changes then
23:48:00Nico_Pbluebrother: very strange
23:48:11bluebrotherand now "ark"
23:48:41*bluebrother checks the genre tag of the file
23:48:56lostlogicNico_P: possibly, I'm trying to figure out which line, but hafta recompile with -save-temps so I can map asm to lines without breaking my head
23:49:04Nico_Pbluebrother: sounds like a problematic pointer
23:49:26*jhMikeS is too whacky to document anything he does :p
23:50:23bluebrothernow I got " GPIO_G: %02x" on a different file
23:50:43jhMikeSbluebrother: it's the surprise string feature :)
23:50:48Nico_Pbluebrother: what target? what kind of file?
23:51:10*moos says good night/day and congrates nico again
23:51:11bluebrotherseems to only happen for mp3 −− I haven't got the problem with an ogg file (yet)
23:51:16bluebrothertarget is ipod mini
23:51:22 Quit moos ("Glory to Rockbox")
23:52:55bluebrothertoo bad the track info screen doesn't show the type of the tag
23:53:53bluebrotherI'm somewhat under the impression it could have something to do with id3 versions
23:55:02pixelmadoesn't the default wps display that?
23:55:28bluebrothergood point.
23:57:18 Quit barrywardell ()
23:57:50lostlogicgcc's too smart for me. It knows that if a loop exits because m is null it doesn't hafta check an if m immediately afterwards.
23:58:17Nico_Pbluebrother: I confirm cuesheets work fine. they just aren't preloaded
23:58:45lostlogicNico_P: the data abort is due to dereferencing an invalid memory_handle it seems, but one which is non-null
23:58:52 Quit mf0102 ("Verlassend")
23:59:01Nico_Plostlogic: is this caused by find_handle?

Previous day | Next day