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 2010-04-12

00:03:04 Quit wodz (Quit: Leaving)
00:03:13 Join skx` [0] (
00:03:53linuxstbWatermark: Are you testing? I need to go to bed soon...
00:04:17 Quit anewuser (Ping timeout: 248 seconds)
00:04:18Watermarkhow do i test it? sorry need a refresh
00:04:45linuxstbDo you have a firmware file to upload?
00:04:47dfktlinuxstb, i'm trying it too... having issues installing the tcctool driver
00:05:05Watermarko sure sec
00:05:23linuxstbWatermark: And you'll need to install the tcctool driver for Windows (I'm assuming you're on Windows...)
00:05:29dfktlinuxstb, here's the latest firmware -
00:07:44Watermarkhow do i install it if no Device can be detected?
00:07:58Watermarkwell, can't be detected by Tcctool
00:08:03 Quit ender` (Ping timeout: 276 seconds)
00:08:13Watermarkwhen i try to install nothing happens
00:09:04dfktlinuxstb, just got the driver to install... i had to add the pid&vid.. d'oh
00:09:22Watermarkdfkt how do i do it?
00:09:55dfktyou open tcc.inf and add this line under "devices":
00:09:55dfkt"TCC7901 usb-boot mode"=LIBUSB_DEV, USB\VID_140E&PID_B057
00:09:55linuxstbdfkt: Ah yes... I forgot about that.
00:10:12dfktadd it 3 times, for each subsection
00:11:15linuxstbdfkt: The COWON_S9_FW.BIN looks like the right file to try to upload.
00:11:18dfktWatermark - or just copy that whole thing over:
00:11:29Watermarkyes omg
00:12:00Watermarkomg i love you guys Crys...
00:12:04CIA-5New commit by Buschel (r25600): Tweaking iPod Video battery configuration. Dangerous battery level is latest reached below 3500 mV, discharge curve is optimized for stable runtime ...
00:12:43dfktlinuxstb, you mean after the *fw.bin is uploaded, the other two firmware parts should already be uploaded in normal connect mode?
00:13:24dfkti let Watermark test that, his s9 can't be more broken than it already is... and i don't want to brick my s9 ;)
00:13:34Watermarki rthink all you need to the FW file to upload then all should be good
00:13:55linuxstbWatermark: So Windows is now recognising your S9?
00:14:02linuxstb(in recovery mode)
00:14:15dfktthe *rs.bin file is obviously resources, like fonts and such - no idea what the *ft2.bin is
00:14:31Watermarkyep, brb ok. going to upload the FW via tcctool. refreashing myself here
00:15:33CtcpVersion from Chazz!dabomb69@unaffiliated/dabomb69
00:20:22 Quit Buschel ()
00:24:59Watermarkhm... nothing seems to be happening
00:25:56dfktare you in recovery mode? is the s9 recognized in windows device manager?
00:26:28Watermark yes, i was able to use tcctool but the device shuts off after patching the FW File
00:26:48dfktdid your battery run dry when you bricked it? ;)
00:27:08linuxstbHave you found any instructions for recovering a Cowon D2 with tcctool? This could be similar (and I don't know that process).
00:27:19Watermarkhmmm... maybe
00:27:41Watermarko i'm an expert on that teehee
00:27:48linuxstbAlso, maybe you need to hold the "on" button when you upload the firmware - so it detects it as being pressed when it runs.
00:27:56Watermarki was actually using one of my guides as a refresh
00:28:40linuxstbOr it could mean the "sdcfg" value I guessed was wrong. But there's no more I can do now - time for sleep... I'll read the IRC logs tomorrow.
00:29:05 Join halmi [0] (~Miranda@
00:29:07dfktthanks, linuxstb
00:29:10Watermarkok thanks linuxstb
00:30:17 Quit Schmogel (Quit: Miranda IM! Smaller, Faster, Easier.
00:31:01dfktWatermark, so did oyu see an "updating" screen after you got the firmware on the s9?
00:31:18dfkti assume you have to push the rest hole for the update to initialize, iirc
00:31:25dfktto get the player out of recovery
00:31:35Watermarkas before the S9 shuts off when patching is done
00:32:33dfktshuts off? you mean, it isn't visible anymore in the device manager?
00:32:49dfktsince there is no other clue of activity when it's in recovery mode
00:33:04Watermarkhold on
00:35:16Watermarkyeah, the device isn't listed after patching. like it disconnect
00:36:25dfktand you pushed the reset hole?
00:37:50Watermarkthe only thing that does is lets me reconnect the S9 in Boot Mode. when i try to turn the S9 on in Boot or regular nothing happens
00:42:41 Join Blue_Dude [0] (
00:43:42 Quit linuxstb (Ping timeout: 258 seconds)
00:44:01dfktthen it's probably best waiting for linuxstb to come back... i'm off to bed as well - good luck!
00:44:04 Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
00:44:33 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
00:44:58 Quit BHSPitLappy (Ping timeout: 276 seconds)
00:46:32 Quit petur (Remote host closed the connection)
00:46:57 Quit EKulabuhov (Ping timeout: 248 seconds)
00:47:40 Join Strife1989 [0] (
00:49:17Watermarkalright thanks
00:50:14 Join nima [0] (
00:51:17 Quit Strife1989 (Read error: Connection reset by peer)
00:51:38 Quit Strife89 (Read error: Connection reset by peer)
00:51:40 Join Strife1989 [0] (
00:53:42 Quit nimak (Ping timeout: 276 seconds)
00:54:32 Quit Watermark (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
00:58:12 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
01:06:03***Saving seen data "./dancer.seen"
01:09:21 Join Watermark [0] (
01:12:55 Quit Watermark (Client Quit)
01:15:33 Quit piotrekm (Quit: piotrekm)
01:17:03 Quit Tuplis (Ping timeout: 245 seconds)
01:17:20 Join Tuplis [0] (
01:20:22 Join guymann [0] (
01:28:04 Quit efyx (Quit: Quitte)
01:32:00 Quit halmi (Quit: halmi)
01:35:23 Quit xiainx (Ping timeout: 245 seconds)
01:37:31 Join Watermark [0] (
01:38:02Watermarklinuxstb you still here?
01:39:57Watermarknvm, we will see what you can do to further fix the issue (Issue: after patching the Device disconnects from PC)
01:41:09 Quit Blue_Dude (Read error: Connection reset by peer)
01:41:10S_a_i_n_tHe went to bed, he'll read the logs.
01:41:21Watermarkroger that
01:41:47 Join arbingordon [0] (~w@unaffiliated/arbingordon)
01:42:49Watermarkbtw linuxstb, i will be on close to 3PM CDT
01:43:07Watermarksee you then and again Thanks!
01:43:16 Quit Watermark (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
01:45:07 Join Spaceghost [0] (libertad@unaffiliated/spaceghost)
01:45:39Spaceghostwith a ipod rockboxed, is possible use the screen for show anything that I want when the ipod is connect to the PC?
01:46:02Spaceghostshow what music I am playing on the computer, for example
01:46:50Spaceghostthe CPU temp, CPU use, RAM use, etc
01:47:14Spaceghostwith a good dock, that is very interesting
01:48:50 Quit DataGhost (Ping timeout: 264 seconds)
01:57:26 Quit krabador (Read error: Connection reset by peer)
02:08:36S_a_i_n_tWhat's with the .sbs flashing when the database inits?
02:09:01 Nick fxb is now known as fxb__ (
02:09:06 Join halmi [0] (~halmi@
02:09:41 Quit toffe82 (Read error: Connection reset by peer)
02:10:04 Join halmi_ [0] (~halmi@
02:10:37 Quit halmi_ (Client Quit)
02:10:49 Join halmi_ [0] (~halmi@
02:11:05 Quit Spaceghost (Ping timeout: 258 seconds)
02:11:18 Quit halmi_ (Client Quit)
02:13:51 Quit halmi (Ping timeout: 268 seconds)
02:14:59 Join IiLuminated [0] (iiluminate@
02:15:08 Join halmi [0] (~halmi@
02:15:32 Part IiLuminated
02:15:38 Quit halmi (Client Quit)
02:28:33 Quit adnyxo (Remote host closed the connection)
02:42:03 Quit kugel (Remote host closed the connection)
02:48:02 Join hearit [0] (~asdd@
02:55:11 Quit GeekShadow (Read error: Connection reset by peer)
03:06:04***Saving seen data "./dancer.seen"
03:07:09 Join xiainx [0] (
03:17:22 Join [1]Lynx_ [0] (~Lynx@
03:19:13 Quit Lynx_ (Ping timeout: 268 seconds)
03:19:13 Nick [1]Lynx_ is now known as Lynx_ (~Lynx@
03:29:32 Join [1]Lynx_ [0] (~Lynx@
03:32:41 Quit Lynx_ (Ping timeout: 260 seconds)
03:32:41 Nick [1]Lynx_ is now known as Lynx_ (~Lynx@
03:37:56 Nick Strife1989 is now known as Strife89 (
03:39:47 Join FlynDice [0] (
03:41:26 Join hebz0rl_ [0] (
03:44:54 Quit katyl (Quit: Ex-Chat)
03:45:09 Join mikroflops_ [0] (
03:45:13 Quit hebz0rl (Ping timeout: 264 seconds)
03:49:50 Quit mikroflops (Ping timeout: 276 seconds)
03:50:49Strife89A minor but perhaps-difficult(?)-to-impliment suggestion:
03:51:08Strife89Two SBSes: one for the Main Menu, on for everywhere else.
03:51:25JdGordon_you can do that anyway
03:51:37S_a_i_n_tyou can?
03:51:40Strife89... You can?
03:51:58JdGordon_as long as the ui viewport doesnt move you can
03:52:09JdGordon_^ is on my todo list
03:52:21Strife89As for why: If you start digging into menus, having a lot of other info on the screen can become frustrating.
03:53:11S_a_i_n_tJdGordon_: How do you inplement these two .sbs? I mean "use" rather than implement.
03:53:21S_a_i_n_tI've thought about it, but couldn;t work out how.
03:53:29JdGordon_with the %cs and %mp tags
03:54:25S_a_i_n_tOh...right, I gotcha. One .sbs, but conditional to screens.
03:55:19S_a_i_n_t<pedantic>well, that's not *really* two .SBS's.</pedantic>
03:55:26*Strife89 can only imagine how complicated such an SBS will look.
03:55:36Strife89In the source, I mean.
03:55:52S_a_i_n_tStrife89: not very.
03:57:11JdGordon_it is the only way
03:57:11S_a_i_n_tThe .sbs file itself wouldn't be too much more complicated actually.
03:57:20S_a_i_n_tIt's given me ideas lol.
03:58:35*S_a_i_n_t still can't believe how much skin buffer he can get back by using %pv|bitmap.bmp|
03:59:17S_a_i_n_tit's truly awesome, some of my themes still need to have bitmap strip volume, but that doesn't mean I can't make new ones!
04:00:22 Quit linuxstb (Ping timeout: 264 seconds)
04:03:22JdGordon_S_a_i_n_t: can you please do a patch for all the cabbies?
04:05:16S_a_i_n_tyep. I guess so.
04:05:19 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
04:05:37S_a_i_n_tmay take me a while, but I can get it done soon-*ish*
04:05:41JdGordon_hmm, I think i need to pix it so it doesnt eat the rest of the line, so it can go in conditionals
04:06:12S_a_i_n_tit doesn't need to be a conditional, does it?
04:06:26S_a_i_n_tit can use %?mv to call it
04:06:36S_a_i_n_t(in a viewport)
04:06:46S_a_i_n_tthat's what I'm doing fine.
04:07:00S_a_i_n_tmv (is volume changing yes/no
04:07:29JdGordon_na, I still want to add the tag for "is volume above 0?"
04:07:55S_a_i_n_tI can't think of many other reasons you'd want volume to be conditional.
04:08:02S_a_i_n_tOh, right.
04:09:09S_a_i_n_tJd, wait. How can I patch the cabbies with this? remove the volume .bmp that exists already?
04:09:39S_a_i_n_tjust use a small "volumebar" in its place?
04:09:43 Quit TheSeven (Disconnected by services)
04:09:56 Join The_Seven [0] (~theseven@rockbox/developer/TheSeven)
04:10:06 Nick The_Seven is now known as TheSeven (~theseven@rockbox/developer/TheSeven)
04:10:07Strife89Y'know, why does the Reversi plugin start you off against a human opponent?
04:10:10S_a_i_n_tGotcha..but some people may not like that.
04:10:21Strife89Wouldn't one rather play against the CPU?
04:10:35JdGordon_once the "clipping" tag is added, noone should be able to tell the differebnce
04:11:31S_a_i_n_tIt doesn't need it, I can just colour the end of the .bmp where it (approximately) clips
04:12:02S_a_i_n_tgo, yellow, orange (just about clipping), red (definitely clipping)
04:12:12JdGordon_or that
04:12:17S_a_i_n_tSame effect, no additional tag
04:12:24JdGordon_although the area changes based on the target
04:12:29S_a_i_n_tI'll get to work.
04:12:53S_a_i_n_tYeah, it'll be approximate...but it was anyway really,
04:13:02S_a_i_n_tthe only value that wasn;t was 0db
04:13:47JdGordon_yeah but old style is <mute|......range....|0dB|>0>
04:14:12S_a_i_n_tHmmm, yeah. this won't have mute.
04:14:35S_a_i_n_twell, mute will be "there is no volume icon"
04:16:15 Quit mirak_ (Quit: Ex-Chat)
04:17:09 Quit linuxstb (Read error: Operation timed out)
04:17:34 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
04:19:47JdGordon_mute is done with a silhouette of the bar on the mackground
04:19:56JdGordon_hehe mackground
04:20:36S_a_i_n_tmy theory of showing clipping with the very end of the .bmp won't work for volume-112x64x1 for instance
04:20:48S_a_i_n_tas the "clip" indicator is at the front of the .bmp
04:21:08S_a_i_n_tI'll do the targets I can, and we'll go from there.
04:25:31 Join Barahir_ [0] (
04:26:19 Join Rob2223 [0] (
04:28:48 Quit linuxstb (Ping timeout: 264 seconds)
04:29:16S_a_i_n_tHmmmmm.I think we need a better way to represent clipping.
04:29:31S_a_i_n_tI could only ever get it *vaguely* approximate.
04:29:38 Quit Barahir (Ping timeout: 276 seconds)
04:29:56 Quit Rob2222 (Ping timeout: 265 seconds)
04:30:14S_a_i_n_tFor the target I'm looking at now, when clipping starts would *technically* be 3dB too "early"
04:30:37JdGordon_a new tag ad fix it so it doesnt eat the line
04:31:00S_a_i_n_tSo, I have to wait untill then?
04:31:19S_a_i_n_twell, I guess so, so I know how it works, so I know what .bmps to make
04:31:57S_a_i_n_tor will it just be 2 bmps? 1 for x~0, the other for 0+?
04:32:15S_a_i_n_tif the latter, I can still start working on it now.
04:32:26S_a_i_n_tand, it'd be a LOT easier.
04:32:49 Quit Strife89 (Quit: Sleep.)
04:33:02JdGordon_yeah, just 2 bmps
04:33:26 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
04:33:45S_a_i_n_tSweet, means I can just use the last two cells from the original volume strips, and noone will notice anything ;D
04:33:54JdGordon_how bad does %?pv<|%pv|bmp|...||0dB|%xdf> look?
04:34:12S_a_i_n_tfine, it's how I imagined it in my head.
04:34:32S_a_i_n_tbut, does that mean the other way of doing it won't work anymore? :'(
04:35:02JdGordon_it would still work, the midde bit is only 1 so it is always used
04:35:32S_a_i_n_tso, they'll both work, min to max *and* min to 0db to greater than 0dB?
04:39:49 Quit linuxstb (Ping timeout: 246 seconds)
04:42:30 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
04:50:24 Join Rondom_ [0] (
04:53:29 Quit linuxstb (Read error: Operation timed out)
04:53:57 Join linuxstb [0] (
04:53:58 Quit linuxstb (Changing host)
04:53:58 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
04:54:01 Quit Rondom (Ping timeout: 248 seconds)
04:59:27 Quit linuxstb (Ping timeout: 252 seconds)
04:59:45S_a_i_n_tOk, sweet, bitmaps for all the colour targets are done.
05:00:06S_a_i_n_tJust need to wait and find out the final syntax, then I can start on the WPS files.
05:00:49S_a_i_n_tI've gone with "volume-240x320x16" and "volume-clipping-240x320x16" etc.
05:01:30 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
05:01:42LloreanDo we need new bitmaps if the old syntax isn't discontinued?
05:02:22S_a_i_n_tbecause it uses one bitmap like the playbar does.
05:02:30LloreanThat's not a reason.
05:02:30S_a_i_n_tnot a cell from a bitmap strip.
05:02:34LloreanThat's an explanation of what it does.
05:02:49S_a_i_n_tIts a reason if you understand it.
05:02:54LloreanNo, it's not.
05:03:44S_a_i_n_tdoes a playbar use bitmap strips?
05:03:48S_a_i_n_tsame thing.
05:03:56 Join markun_ [0] (
05:03:58 Quit markun_ (Changing host)
05:03:58 Join markun_ [0] (~markun@rockbox/developer/markun)
05:04:09LloreanWhy does it need to be like the progress bar (learn your terms)
05:04:28LloreanI'm asking for reasons, not just "what it does"
05:04:31LloreanI understand what it does.
05:04:34 Quit markun (Ping timeout: 276 seconds)
05:04:47S_a_i_n_tbecause that's an option just committed, jd asked me to do something, I'm doing it.
05:05:07LloreanAgain, "it's a new option, so we should use it" isn't a reason.
05:05:09LloreanWhat's the benefit?
05:05:26LloreanWhat do we gain by replacing the existing tag in WPSes with the new version?
05:05:35S_a_i_n_tless skin buffer realty
05:05:49S_a_i_n_ta LOT less in some cases.
05:06:08***Saving seen data "./dancer.seen"
05:06:43LloreanDoes the amount of RAM allocated decrease?
05:06:53LloreanOr does it simply decrease the total amount used within the allocation we already give it?
05:07:04 Quit CaptainKewl (Remote host closed the connection)
05:07:04S_a_i_n_tdoes it really matter?
05:07:08LloreanYes, actually
05:07:24LloreanYou can't use allocated WPS RAM for other things even if the WPS isn't using all of it
05:07:32S_a_i_n_tOk...then argue it with the author, not I.
05:07:33LloreanSo if the amount of RAM allocated doesn't decrease, it doesn't actually save anything
05:07:38S_a_i_n_tactually *discuss* it.
05:08:10LloreanYou mean, like asking for reasons to do it and expecting people to actually be able to give me information such as "benefits" rather than reiterating what it does, or making things up that they don't even know about? That sort of discussion?
05:08:44S_a_i_n_tSome people may not think aesthetics matter, others do, those that don't aren't losing anything.
05:09:19LloreanWell, if people are to be able to learn from cabbieV2 too, having an example of both multiple images (the volume indicator) and a single image that is progressively drawn (progress bar) provides the wides benefit to people looking for examples.
05:09:55S_a_i_n_tYou just like to shake up things you dislike right? its totally uncool.
05:10:09LloreanWhat's the aesthetic gain, exactly? JdGordon_ said people wouldn't be able to tell the difference. If you're making an exact replacement, surely you can't argue that it's also an aesthetic gain, so either there's a visible difference (which you can then call a "reason") or it can't be told apart.
05:11:01LloreanS_a_i_n_t: I notice that you're making personal attacks against me already, rather than discussing the idea of whether or not it decreases the usefulness of it as a learning tool.
05:11:15LloreanScore one point for the person who suggested *I* should actually try discussion.
05:11:36S_a_i_n_tNot an attack, merely an observation, based on conduct past and present.
05:11:56LloreanYes, one that doesn't belong in this discussion.
05:12:01LloreanUnless it's somehow a part of the feature.
05:12:32LloreanSo, do you have answers for *any* of the questions I've presented? If you want to complain about my behaviour rather than actually discussing the idea, I suggest it be taken to -community.
05:12:38LloreanAs it stands, I'd like to know what we're gaining from this.
05:12:44S_a_i_n_tDid I or did I not ask you to refer you "questions" to the author?
05:12:50S_a_i_n_tyet you still continue.
05:13:10LloreanYou claimed something about aesthetics *after* that
05:13:13LloreanWhich I then asked about
05:13:41LloreanI assumed you were offering more information on that aspect - why otherwise even say it?
05:21:45 Quit linuxstb (Ping timeout: 245 seconds)
05:25:23 Join Adubbb [0] (~Aldubuc@
05:27:03 Part Adubbb
05:27:30 Join JdGordon [0] (~7bf38c1f@gateway/web/freenode/x-jruphgujmmcmtzst)
05:27:38JdGordonyou are both idiots!
05:27:49LloreanJdGordon: I'm just asking what we gain.
05:28:17LloreanIf it's exactly the same, I don't think we need to change cabbiev2, but if we gain something, we certainly could.
05:28:26 Join Adubb [0] (~Aldubuc@
05:28:37 Part Adubb
05:28:40JdGordon1) much simpler wps code, 2) skin buffer back so it can be used for other things, 3) smoother bar drawing
05:29:13LloreanJdGordon: I don't agree with 1) as long as the old syntax still exists. 2) I already asked my question about (if we really get it back from the allocation, I'm all for it). 3) wasn't told to me.
05:29:32LloreanIs the current bar on many targets inaccurate? I imagined it was small enough on most targets that we'd have enough bitmaps to be precise.
05:30:00LloreanI know that currently I have several hundred K allocated for my WPS that is unused according to rockbox info
05:30:28JdGordon%?pv<%xdCa|%xdCb|%xdCc|%xdCd|%xdCe|%xdCf|%xdCg|%xdCh|%xdCi|%xdCj> is nicer than %pv|volume.bmp|-|-|-|-| is it?
05:30:35LloreanNo, it's not nicer.
05:30:39LloreanI didn't say it was.
05:31:06LloreanMy point was that as long as we allow the old syntax, it's better to have it in cabbiev2 is there's not an aesthetic difference on the actual screen, because it's the one people will have a harder time understanding.
05:31:37LloreanBut if there's a precision problem with the old code, then an upgrade might make more sense.
05:31:51LloreanAlso, of course, if we got rid of the old syntax.
05:32:16LloreanBut I'm not sure that's feasible since it servers for nonlinear volume bars too
05:32:33 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
05:33:10JdGordonfrankly I dont care, I think cabbie should show good WPS code, I asked S_a_i_n_t if he would do it, he said maybe, thats it
05:33:57LloreanI think that, where it doesn't harm other things, it should be a decent learning tool.
05:34:02JdGordonand on that, I would also really like to force cabbie on all targets to use viewports
05:34:26LloreanIs there some reason it doesn't need to on some targets?
05:34:49LloreanI guess on Portrait targets it can get away without it...
05:35:03LloreanWhat about the newer cabbie versions from the theme site that take advantage of SBS features and such?
05:35:10LloreanMaybe it's time for an official cabbie version 3.
05:35:16JdGordonok, any target which cabbie is filled with %?C<>'s NEED to be converted
05:35:49LloreanSounds like a good idea.
05:39:09saratogagoogle has kicked some money to the Tremorlo guy
05:41:16 Quit Horscht (Quit: Verlassend)
05:41:21 Join anewuser [0] (anewuser@unaffiliated/anewuser)
05:41:47S_a_i_n_tJdGordon: I'm all set when we have a definitive syntax to work with. Just post an example line somewhere for me.
05:42:04 Join Spaceghost [0] (irrsi@unaffiliated/spaceghost)
05:42:34JdGordonany ideas for the letters to use for "is above 0?"
05:42:59JdGordonI think a new token is much more readable than putting a %pv|...| insdie a %?pv<>
05:43:05LloreanIs $vd taken?
05:43:09LloreanEr %
05:43:30LloreanAh, the V stuff seems to all be viewports
05:43:36JdGordondon't think so
05:43:42JdGordonyeah V is viewport stuff
05:43:45LloreanI'd say "vd" for "volume danger" or something
05:43:48JdGordonbut we only have so many letters
05:44:20JdGordonI can do %vpd or something
05:44:26JdGordonumm %pvd
05:44:41S_a_i_n_tJdGordon: Just go with %xda|%xdB|etc.?
05:44:49S_a_i_n_t*for "above 0dB"
05:45:01JdGordonno, a single image for above 0
05:45:05S_a_i_n_toh, shit...wait.
05:45:20LloreanJdGordon: Would it be possible to concatenate two images?
05:45:35 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
05:45:38JdGordonyou mean overlay?
05:45:40LloreanSo you could have %pv|muteimg|mute+1 through 0db| danger.bmp|
05:45:52LloreanNo, basically, take two images and draw one until it's wholly there, then start drawing the other
05:45:58LloreanFrom the end of the first
05:46:03S_a_i_n_tjd, we only need one letter for "is volume above 0dB" right?
05:46:15JdGordonthat is a seperate feature which we've talked about
05:46:41JdGordonthere is a patch for it, but I dont tihnk it is needed because doing the volume bar with a single image looks better
05:46:54JdGordonS_a_i_n_t: tags are 2 letters
05:47:07S_a_i_n_tbut, doesn't allow for accurate "am I clipping" unfortunately.
05:47:27S_a_i_n_t*using only one bitmap.
05:47:33S_a_i_n_tthis is why we need to.
05:47:55LloreanJdGordon: Could it be |mute|scaling image|danger.bmp ?
05:48:15S_a_i_n_tthere is no mute IIUC
05:48:22LloreanThere is mute.
05:48:28JdGordonwith the current tags it is mute|scaling|0|danger but that means %pv inside %?pv
05:48:29LloreanAfter the lowest volume level it mutes the DAC
05:48:59S_a_i_n_tnot with this implementation there isn't
05:49:19LloreanS_a_i_n_t: There's a difference between "there is no mute" and "we don't display mute with the current implementation"
05:49:23LloreanPlease, say what you mean. I can't read minds.
05:49:49LloreanJdGordon: Maybe we should just go with the nested tag for the moment until someone has a better idea, just so it can get in use?
05:49:56LloreanIt does offer some flexibility that way, at least
05:50:28JdGordonthat snippet wont work right now because the long type %pv eats the rest of the line (simple fix)
05:50:42 Quit panni_ (Quit: ( :: NoNameScript 3.81 :: ))
05:50:44JdGordonand I wouldnt remove that option if another tag was added
05:50:55JdGordonbut I think it is a bit awkward
05:51:13LloreanWell, it would be nice if there was an actual mute image as well
05:51:44 Quit kramer3d (Quit: This computer has gone to sleep)
05:51:52S_a_i_n_tI assume mute will be the empty recess in the volume area, well need to add it to some backdrops
05:52:13LloreanYou could have %sa or %sv with a basic |mute|!mute|danger condition set.
05:52:46LloreanS_a_i_n_t: Often there's a grayed out volume behind the volume bar showing where it will fill in.
05:52:55LloreanSo I don't think you could add mute to the background without removing that.
05:53:02JdGordonI don't think any extra tags should do more than just be a bool for if the volume is above 0 or not
05:53:11S_a_i_n_tI know, but I don;t think all the targets have this.
05:53:16LloreanJdGordon: What about mute, then?
05:53:17S_a_i_n_tBut I could be wrong.
05:53:27JdGordonis mute such a big deal?
05:53:36S_a_i_n_tIMO, no.
05:53:44JdGordonyou also dont tihnk >0 is :D
05:53:54LloreanI think >0 is a lot more important than mute
05:54:00LloreanBut a mute indicator has revealed a bug from time to time
05:54:10LloreanWhen you see that it's supposed to be muted, and you still hear audio, you know something's up.
05:54:31LloreanGenerally speaking, you're going to pause rather than mute anyway.
05:55:03S_a_i_n_tIf we had a "mute" button, I'd say it was important.
05:55:09S_a_i_n_tbut, seeing as we don't.
05:55:35*JdGordon wonders about allowing a different seperator style for when we are inside a conditional
05:56:02LloreanJdGordon: Why not %bv as a new tag with |mute|scaling bar|danger ?
05:56:13LloreanRather than trying to make it coexist with %pv etc?
05:56:55S_a_i_n_tpersonally, I think %pv still makes sense. The systax that follows is enough to tell the two %pv's apart.
05:57:16JdGordonbecause unless the scaling bar part of your suggestion is another tag, it is to restricted
05:58:43LloreanCouldn't you make it so if the first/third parts were empty it works like your %pv does, but if they were filled in, it shows the respective images at mute / >0, without having to dance around the fact that %pv could *theoretically* have a single bitmap in the middle?
05:59:30JdGordonI dont like forcing anything, what you suggest forces the bitmap, but the themer might want something compeltly different
05:59:39JdGordonso it really just duplicates the existing pv tag
06:00:04LloreanI don't understand
06:00:06LloreanForces which bitmap?
06:00:31LloreanI was just suggesting instead of the new tag being another limited variant of %pv, give it more flexibility by making it its own tag.
06:00:33JdGordonmaybe i dont understand you
06:00:35LloreanThen leave the old %pv alone.
06:01:02LloreanBasically, instead of two different %pvs, have the old %pv, and a new %bv which takes the form %bv<mute| scaling bar image | danger>
06:01:11LloreanWhere if mute and danger are left empty, it just uses the scaling bar.
06:01:34JdGordonand the image is what? any wps code? or a bmp filename?
06:02:14LloreanI'm confused.
06:02:25LloreanDoesn't %pv<volume.bmp> require an image too?
06:02:41S_a_i_n_tof course.
06:02:41LloreanEr <volume.bmp|-|-|-|-> or whatever
06:02:56LloreanI'm basically proposing use that .bmp in a new tag, rather than a variant of the old tag.
06:03:00LloreanSo that it's more explicit for users.
06:03:07LloreanAnd so the theme code can still be rather clean
06:04:27S_a_i_n_tI still think it makes more sense to have just the one tage for "draw volume" and let the syntax that follows it determine how its drawn.
06:04:42S_a_i_n_trather than having two tags with *similar* outcomes.'
06:05:15LloreanWe've got lots of different tags for very similar things
06:05:35Llorean%ff / %fk, %fm / %fn, etc.
06:05:40JdGordonLlorean: the problem is those extra params to say where to draw the bmp
06:05:54JdGordonanything we do is going to be messy
06:05:58LloreanAh, true
06:06:15LloreanI was just thinking it might be less messy if the progress bar-like version had its own tag, so someone could tell at a glance what they're trying to decipher
06:06:35S_a_i_n_tthe syntax should indicate that.
06:06:41JdGordonit is obvious which volume display is being used by the code
06:07:31LloreanI'm confused.
06:07:31JdGordonand I'm liking %?pv<xdaA|%pv,volume.bmp,-,-,-,-,|%xdaB|%xdaC> if you want full control
06:07:41LloreanWhat does the new tag look like exactly?
06:08:05S_a_i_n_tJdGordon: btw, I think using "," instead of "|" makes it cleaner, better than the mess that %Sx can look like
06:08:08LloreanI mean the old tag is %?pv<Mute|N entries|0 dB|Above 0 dB>
06:08:16LloreanHow do you distinguish the new tag, exactly?
06:08:21LloreanThe presence of a filename as the first entry?
06:08:22S_a_i_n_tthese tags systaxes are vaguely similar.
06:08:46JdGordonLlorean: %?pv<> is a conditional, %pv is a number, %pv|||||| is a bar
06:09:39LloreanWill it draw without a .bmp as our old progress bar code?
06:09:59JdGordonit is all the same code, just wired to different values :)
06:10:28LloreanI guess it makes sense then.
06:10:34LloreanDo we *need* a >0 conditional then?
06:10:44LloreanThe basic %?pv<|||> conditional seems to do it.
06:10:49LloreanIf someone _must_ have that functionality.
06:10:53JdGordonthis is the question
06:10:54LloreanAnd it offers mute as well.
06:11:04LloreanI don't think we need to add yet another tag for something that's already done.
06:11:46LloreanIt's going to look almost equally ugly with the two-case as with the four-case conditional
06:11:49LloreanOnly very slightly better.
06:13:14*JdGordon is going to fix it so the param delimiter *inside* conditionals can be user defined
06:13:30JdGordonor would that be too flexible?
06:13:38JdGordonmaybe just allow ,./ ?
06:13:51JdGordonor ' " :
06:13:51LloreanI'm not following.
06:13:53S_a_i_n_twait, noob that up for me.
06:14:02S_a_i_n_tMe neither lol
06:14:16LloreanOh, wait
06:14:23LloreanYou mean allowing the use of something other than | in conditionals
06:14:27*Llorean somehow didn't see the /me line
06:14:28JdGordon%?pv<xdaA|%pv,volume.bmp,-,-,-,-,|%xdaB|%xdaC> <- readable %?pv<mute|%pv|volume.bmp|-|-|-|-||0db|dander> <- de fuck?
06:15:00LloreanMaybe allow conditionals to "eat" line breaks, so you can nest them across multiple lines for readability?
06:15:22JdGordonthat would probably be a nice addition also, but not really related
06:15:23LloreanAllowing multiple delimiter choices just seems like a recipe for extra confusion as different WPS authors use different styles.
06:15:33LloreanWell, they both help readability in a situation like that.
06:16:10S_a_i_n_tIMO %?pv<xdaA|%pv,volume.bmp,-,-,-,-,|%xdaB|%xdaC> is the winner of the two.
06:17:07S_a_i_n_tbut, having a bitmap for 0dB is slightly hard to do.
06:17:44Llorean %pv|volumte.bmp|-|-|-|-||
06:17:44Llorean 0db|
06:17:44DBUGEnqueued KICK Llorean
06:17:44Llorean etc
06:17:46Lloreanisn't too bad, though
06:18:06 Quit linuxstb (Ping timeout: 252 seconds)
06:18:13LloreanS_a_i_n_t: I'd imagine for 0db you'd just re-use the full volume bar image.
06:18:13S_a_i_n_tstill, if the rest is progressbar style, how is 0 and >o represented?
06:18:27LloreanThe volume bar shouldn't appear "full" at -1db
06:18:31JdGordondifferent colour
06:18:31LloreanSo that'd be the only way to see it.
06:18:43JdGordonand the !
06:18:47LloreanFor >0 a different color, or even just something displaying to the side.
06:19:00LloreanLike you could show the full volume bar, but put a warning triangle to the right of it.
06:19:04LloreanIt's up to the user.
06:19:15LloreanCurrent targets in cabbiev2 just turn the volume indicator red (on color targets, obviously)
06:20:32S_a_i_n_tthe way I was thinking it was going to work (which is easiest when it comes to reducing .bmp's which is the goal here is it not?) is, two bmps drawn playbar style, image one is drawn from mute to 0dB, image two takes over from >0dB to max.
06:20:35S_a_i_n_tyou dig me?
06:21:27S_a_i_n_tI was planning on using the last two images from the current .bmp strips for this.
06:21:46S_a_i_n_tI think this works best (in theory), but how to implement it.
06:21:47JdGordon0 should be full or almost full
06:21:55JdGordonfor cabbie, 0 SHOULD be full
06:22:08JdGordonfull yellow
06:22:10LloreanFor cabbie currently, 0d is the only time the "full" image is shown
06:22:20Llorean-1 gets a nearly full, and +1 gets red.
06:22:31S_a_i_n_t+1 to +6
06:22:40Llorean+1 to +whatever
06:22:56LloreanSome targets stop at 0, others go up to +12 I think
06:23:26 Quit JdGordon (Quit: Page closed)
06:23:51S_a_i_n_tI'd still like to (as its moddelled from the progress bar) be able to display incremental change past 0dB
06:24:17 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
06:24:23LloreanWhen implementing the old way (mute|increments|0|danger) we decided that wasn't too important.
06:24:25S_a_i_n_tthe way it is in current SVN is fine, but the question/difficultty with that is where/how to start drawing "clipping"
06:24:39S_a_i_n_tas, some targets are plaing with one 11px across.
06:24:55LloreanBasically, the volume indicator isn't too useful past 0db other than letting you know in advance that if you unpause it's going to be very loud.
06:25:15LloreanS_a_i_n_t: Look at how it's done currently...
06:25:21S_a_i_n_tstill, if its PB styled, it *should* still advance past 0db
06:25:21LloreanThe whole image is overwritten with a different one
06:25:46LloreanIt just needs to be indicated, there are ways to do this besides advancing.
06:26:14CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
06:26:14*jhMikeS has an idea for all audio settings: all settings range 0..n-1, with n being the number of levels. All can be translated easily to dB or other units and back again. Only one value is required in the driver. Any increment or number of decimal places can be represented anywhere, or rounded to nearest int if so desired.
06:26:23S_a_i_n_tbut its nice to know "how far am I above 0db" instead of "I'm just past 0db"
06:26:49jhMikeS* of course this is internal, not user-visible, unless it increases accuracy of display.
06:27:41S_a_i_n_tI dont like the idea that you'd hear an audible increase in volume, yet stop seeing any visual reference of this past 0dB
06:27:47S_a_i_n_tthat's my quark with it.
06:28:11LloreanS_a_i_n_t: In a lot of cases you won't hear an audible increase in volume so much as "more and more distortion at a similar loudness"
06:28:13 Quit n17ikh (Ping timeout: 240 seconds)
06:28:32jhMikeS"quark? you mean qualm?
06:29:05S_a_i_n_tIt just seems to break with the idea of drawing it as a progressbar, if we just stop doing incremental change at 0dB, we may as well not bother with it at all.
06:29:06jhMikeSperhaps it was a "strange quark". :P
06:29:39S_a_i_n_tthe point of progressbar it to show min to max value IMO
06:29:58S_a_i_n_tregardless if that happens to exceed 0dB
06:30:16LloreanWe had someone saying that last time.
06:30:33LloreanBasically, 0db is *a* upper bound
06:30:46S_a_i_n_twe need to note where it is exceeded visually of cource, but the idea that we stop drawing increments past 0dB seems pointless to me.
06:30:52LloreanYou're fixating on the numerical maximum, rather than the realistic upper bound.
06:30:56S_a_i_n_twe may as well use bitmapstrips for that.
06:31:09*jhMikeS agress: bar full when no further increment is possible, bar empty when no further decrement is possible. (if he properly understands what's being discussed).
06:31:11LloreanWhat does how we draw it have to do with what we limit it at?
06:31:18S_a_i_n_tthat "someone" was me ;)
06:31:35S_a_i_n_tand as per current SVN, I won that battle (sarcasm)
06:31:57 Quit hearit (Read error: Connection reset by peer)
06:31:58LloreanWhy would bitmap strips be better for the 89 steps below 0db, but we need a progress bar if we're gonna to 95 steps?
06:32:31LloreanBasically, "100% volume" is 0db, and above that would best be represented but >100%
06:32:35LloreanSo a "full bar" is 0db.
06:32:47S_a_i_n_twell, IMO a progressbar should show ever increment it can, not just stop at a value it is capable of exceeding.
06:32:52S_a_i_n_teven if you shouldn;t
06:32:57 Quit linuxstb (Ping timeout: 276 seconds)
06:33:04S_a_i_n_tusers know that "full" volume is a bad idea.
06:33:17 Join n17ikh [0] (
06:33:21LloreanMany of them don't know of the dB scale or what exceeding it can mean at all
06:33:46S_a_i_n_texactly, but *everyone* knows that max volume is bad.
06:33:56jhMikeSyou can exceed the dB scale?
06:34:06S_a_i_n_tso, I don;t see why increments beyond 0dB pose a problem
06:34:10LloreanSorry, what exceeding 0db can mean
06:34:37LloreanS_a_i_n_t: But they're hardly necessary, obviously.
06:34:57LloreanAnd you keep talking about "a progressbar" that goes the whole range.
06:35:08LloreanThere should be a clear visual indicator when 0db is reached.
06:35:27LloreanWhich either means a progress bar image that already indicates where the transition is (difficult to make cross-target compatible) or two progress bars rather than one.
06:35:51S_a_i_n_tIn terms of a "progressbar" I think they are, the reason to have this (initially I thought it was going to be an option, but it seems it's taking over, it's optional *now* and I think it should stay that way) is so people can choose to display the "full" range if they wanted to.
06:36:04S_a_i_n_tactually, I think current svn is fine.
06:36:18LloreanPeople can see the full range by displaying a numeric volume.
06:36:22S_a_i_n_tcommenting out a line for one way or the other in a theme may be an option to consider.
06:36:33LloreanA visual volume is always going to be approximation unless you have as many pixels in width as you have steps.
06:37:50S_a_i_n_tAt the moment, they way I have done it in my theme is this. the last 10% of the volumebar is coloured orange, the final 7% is red.
06:37:55S_a_i_n_tworks fine for me.
06:38:11LloreanThat doesn't tell explicitly where 0db is, does it?
06:38:13S_a_i_n_tbut for smaller .bmp's that is how this whole discussion came about.
06:38:18LloreanNot every 320x240 target has the same range.
06:38:33S_a_i_n_tit can't be represented that way, so we need a seperate image for >0dB
06:39:03S_a_i_n_tLlorean: no, but it gives an idea at least of loud, kinda bad, very bad.
06:39:09S_a_i_n_twhich is all we really need.
06:39:12LloreanS_a_i_n_t: No.
06:39:15Llorean0db is an important position.
06:39:27LloreanKnowing explicitly where 0 is is kinda the important part about all of it
06:39:32S_a_i_n_tsorry: kinda bad was my 0dB
06:39:53LloreanAnother problem splitting it into two bars - if the first part is 90 pixels, and the second 10, what about targets that only go up to 0db?
06:40:06LloreanOr targets with a different ratio (-57 to +6 vs -89 to +6)
06:40:11LloreanStep sizes will vary for the two bars
06:40:34jhMikeSit's all sort of relative anyway. we could just make it so targets have maximum as 0dB without exception, or put 0dB elsewhere on the scale...say, all targets do +6. that's a little deeper change, however.
06:40:37S_a_i_n_tI assume in that case, at 0dB it would show the second "full/past 0dB" image at full.
06:40:39LloreanThe problem is many targets have a 320x240 screen, and many of those have different volume ranges, yet the targets should in most cases share WPSes
06:41:07LloreanS_a_i_n_t: So you'd show a red clipping warning image at 0db where clipping is very unlikely?
06:41:26LloreanAnd that still doesn't address that the number of pixels per step at less than 0db will be different than at >0db depending on target.
06:41:36jhMikeSthis -89 range on the S is done artificially anyway. others could be hooked up to do it though.
06:42:35S_a_i_n_tThe way this is going, it's looking like its headed to "just show from min to max"
06:42:55S_a_i_n_tas it is now in SVN
06:43:15LloreanMax shouldn't be a positive number, though
06:43:20S_a_i_n_tthe exceptions make it too difficult.
06:43:21LloreanOr rather, the maximum value on the bar.
06:43:30S_a_i_n_tand (at the moment) it is optional
06:43:52S_a_i_n_tso, I think a "comment this line and uncomment this line if you want this feature might be aplicable.
06:44:53S_a_i_n_tAt the moment, if themers use it, they know what values it will display, and should be aware of any risks that may come with it.
06:45:42S_a_i_n_tBut as I said earlier, eberyone knows full volume, or even near to it, is a bad idea.
06:46:25S_a_i_n_tThe war against stop at 0dB or stop at Max Vol is never going to be won without someone just making a descision.
06:46:35LloreanThis really has very little to do with "loud is a bad idea"
06:46:50*jhMikeS doubts it really is "everyone". There's always a few of them that just cannot help it.
06:47:03LloreanjhMikeS: There are plenty who've said +6 isn't loud enough and asked if we could make it louder
06:47:09LloreanIt's about a visual indication that you've gone beyond 100% volume
06:47:55S_a_i_n_tand personally, if it were anything other than a progressbar, I'd gladly say "by a;ll means 0dB should be "max", but as it *is* a progressbar, I have to say "it should be min!max volume"
06:48:15jhMikeSI got that much so far. If we go to +6, I think that's all it's got. Beyond that, you need more efficient phones.
06:48:55LloreanS_a_i_n_t: It should go from 0-100%, that's different from "min-max"
06:48:58 Quit hebz0rl_ (Quit: Ex-Chat)
06:48:58jhMikeSthen again, they could just be impaired. I never listen to anything near 0db.
06:50:02S_a_i_n_twell, no, as you'll take 100% as 0dB right? when in my eyes "max" is +whatevr.
06:50:05saratogajhMikeS: the 0dB point isn't really relative, by convention its the maximum level of the amp, with positive values corrisponding to overdriving
06:50:10saratogaso you risk clipping
06:50:15LloreanS_a_i_n_t: 0db is - 100% volume. Say a song is recorded at arbitrary loudness 50. If you add 0db to it, it's still 50 (100% of its original level). If you add 1db to it it's no >100% its original level. If you subtract from it, you're at <100%. Down to mute, which is 0%.
06:50:42Llorean100% is 0db, plain and simple. Nothing added or subracted.
06:50:50jhMikeSsaratoga: is that always the case? I think you're right though. it sounds like it should mean something.
06:50:50S_a_i_n_tbut we're IMO representing minimun/maximun output.
06:50:57S_a_i_n_tnot a dB scale with this.
06:50:59saratogain theory using positive gain values is no better then applying digital gain with the preamp, you will get clipping if you do it too much, but in practice the amp often handles it better then doing it digitally
06:51:05LloreanS_a_i_n_t: No, we're representing a percent of the volume. P V
06:51:31saratogajhMikeS: well thats obviously hardware specific, but afaik we always set 0dB to be max voltage, and if its not, then that would be a bug
06:52:42saratogai bet in practice some DACs have extra headroom built in just to be safe
06:52:53jhMikeSsaratoga: most of my players allow additional gain above 0dB and I've never programmed a driver to limit it to less than the register range.
06:53:16saratogain theory that extra gain is above the maximum output voltage of the DAC
06:53:30saratogaso its only louder if the digital output isn't peak normalized
06:53:56saratogai guess you could pretty easily test this with RMAA though to see if its actually the case
06:54:33saratogaby that i mean only louder without distortion if the digital output isn't peak normalized
06:54:37jhMikeSright. but, a signal up to -6dB could go without clipping if it goes to +6.
06:54:42 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
06:55:12S_a_i_n_tOK, howabout this. We use 2 bitmaps (the third to last, and last cells from each volume bmp strip), the end of the third to last image is 0dB, and beyone that the second image "full red" starts to be displayed?
06:55:41S_a_i_n_tso fullred would be anything past 0dB
06:56:36saratogaisn't that what we do now in cabbie?
06:57:01*jhMikeS thinks the WPS parser and display would handle this best, by marking the "0dB point" somehow in the .wps file. That way it's always correct.
06:57:15S_a_i_n_tnot quite.
06:57:33S_a_i_n_tthe second to last image is 0dB, but its "full yellow"
06:57:43saratogathats what it should be
06:57:51saratogayou're not overdriving until you get past 0dB
06:57:56LloreanIsn't that what cabbie does?
06:58:07S_a_i_n_tI suggested using the third to last, as there would still be room for increments past 0dB to be displayed.
06:58:28S_a_i_n_texactly, that's what I suggested.
06:58:36JdGordon_... still going?! :p
07:00:09 Join notlistening [0] (
07:00:22S_a_i_n_tI think we have the champion...this seems to keep the masses happy, and keeps in line with it being a "progressbar"
07:00:31LloreanS_a_i_n_t: I'm confused.
07:00:40LloreanYou want to keep two extra images for "steps" or ?
07:00:49notlisteningFlynDice you have any progress on your clip+ yet?
07:00:55S_a_i_n_tNo, I want to only use two bitmaps
07:00:57LloreanOr is it two separate progress bars, one on each side of 0db?
07:01:08jhMikeSis the number of images fixed (rather WPS ignorant here)
07:01:09LloreanIf so, remember I asked about step size, and total spacing for players that stop at 0db
07:01:23LloreanYou'll have different step size on the left and right side of 0db
07:01:35LloreanAs well as having completely different use of space on players that stop at 0
07:01:50S_a_i_n_tthere isn't one that works for all targets!
07:02:00LloreanThere's just a lot of complication for trying to visualize the *amount* past 0db.
07:02:12LloreanPlus, if someone needs the exact amount they can use a theme with the number rather than visual bar
07:02:17LloreanThat's kinda why the number exists.
07:02:40JdGordon_jhMikeS: no, the total size of the loaded bmps are though
07:02:43LloreanWhat the visual /graphical version does is needs to indicate approximate loudness, 0db, and the potential for clipping in some manner.
07:03:39jhMikeSLlorean: Different step size is a problem, below vs above? is it really meant to be accurate? I admit I don't think too much of it looking at a bar. It just gives me a vague idea.
07:03:45JdGordon_can everyone remembver that the number of pixels used is tiny comapred to the number of steps so any visualisation is stupidly approximate anyway
07:04:07 Join CharlesNumber2 [0] (
07:04:20LloreanJdGordon_: That's kinda my point. I don't know why we need an approximation for use above 0db in the first place. On a lot of player it's going to be 1 or 2 steps at best anyway.
07:04:36 Quit xiainx (Quit: Good Bye)
07:05:15jhMikeSother than this magic 0dB point, accuracy really isn't necessary
07:05:23LloreanjhMikeS: Well, my point is really that there needs to be a clear indicator when you're at 0, and it should be done in a way to try to take up the same amount of space on the screen independently of the volume range of the target (for those with a max of 0 and those with a max above 0) and that mostly comes out to "-X to 0, with a >0 indicator"
07:05:42LloreanWhich is what Cabbiev2 does now anyway
07:05:55 Join S_a_i_n_t_ [0] (S_a_i_n_t@
07:06:04S_a_i_n_t_bah...connection dropped.
07:06:11***Saving seen data "./dancer.seen"
07:06:38 Quit S_a_i_n_t (Ping timeout: 260 seconds)
07:06:55 Nick S_a_i_n_t_ is now known as S_a_i_n_t (S_a_i_n_t@
07:07:51CharlesNumber2Hello all. I am encoding videos for my 5th generation iPod that runs the latest version of rockbox. When encoding with WinFF at a size of 320x240, the video doesn't play as fast as the audio and skips several frames every about every second. Is there a certain bitrate that the mpeg play will run best with?
07:08:03 Quit linuxstb (Ping timeout: 276 seconds)
07:08:14 Part The-Compiler
07:08:55S_a_i_n_tdoesn't winFF have presets for RB targets?
07:09:09CharlesNumber2yes and when using that preset it skips
07:09:17 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
07:09:32 Join einhirn [0] (
07:10:03S_a_i_n_tLower the framrate?
07:10:04jhMikeSLlorean: yes, that problem is rather sticky. it seems two things are in conflict. personally, I'd like to see some space if volume is left, while other targets wouldn't completely fill if they don't go above 0.
07:10:17LloreanCharlesNumber2: The iPod 5G is basically just not fast enough to do MPEG-2 well, and we can't use the hardware for MPEG-4
07:10:57CharlesNumber2Oh I see, Llorean, is there a better codec for the iPod 5G?
07:11:22S_a_i_n_tThis whole thing is messy (volumeprogressbar), I wish to the Gods on high that one way would "just work" so I can get on and start making the patch up.
07:11:58CharlesNumber2And it seemed to play fine when encoded with the size of an iPod nano screen, which led me to believe that the issue was bitrate
07:12:01S_a_i_n_tThe whole targets the stop at 0dB seems to be the culprit
07:12:07saratogayou should just use the original firmware for vdieo on that player
07:12:17S_a_i_n_tand I'm tempted to say, "just don't use it for those targets"?
07:12:59jhMikeSon S, a single bar step is 12-13 dB, +6 would never register a change anyway.
07:13:00LloreanS_a_i_n_t: Why can't you just do it the way that's worked for cabbiev2 for so long?
07:13:20LloreanI thought the point was "show what we did before, but with finer steps"
07:13:22 Quit CharlesNumber2 (Quit: CGI:IRC (EOF))
07:13:25*jhMikeS spoke of cabbiev2 of course
07:13:45S_a_i_n_tAs I've stated, I have beef with the idea that increments past 0dB aren't displayed.
07:13:52jhMikeSLlorean: I'd welcome at least pixel-level step for sure. It's way too coarse.
07:14:09LloreanS_a_i_n_t: The fact that you have beef with it doesn't change the fact that that's currently how the theme is designed.
07:14:24LloreanImplementing a new feature with the same design, and changing the design, are two different tasks anyway
07:14:37LloreanjhMikeS: This new feature does exactly that, AFAIK
07:14:54LloreanjhMikeS: Basically, it progressively reveals a bitmap based on volume level, just like the bitmap progress bar.
07:15:08S_a_i_n_twell, I didn't necessarily think we were sticking to the original design.
07:15:19LloreanS_a_i_n_t: At the very least they should be presented as two separate patches.
07:15:24S_a_i_n_tas much as we cold sure, but not strictly.
07:15:37Llorean"as much as we could" would mean not change the red display...
07:16:14S_a_i_n_tI want it to display red, it's only the targets that stop at 0dB messing this idea up
07:16:38S_a_i_n_tif it wasn;t for that my past idea of third to last and last cells from the original volume would work fine.
07:16:44LloreanIt only messes the idea up if you insist on adding progress after red.
07:16:48LloreanSomething that isn't there currently
07:17:30S_a_i_n_tprogress after red makes sense, to me, if there *is* progress after red.
07:17:38S_a_i_n_tif there wasn't, then it wouldn't
07:18:13jhMikeSLlorean: ok, I thought so. the problem is the same, multi-image or progressive revealing. :)
07:18:28S_a_i_n_tjhMikeS: basically, yes.
07:18:29LloreanS_a_i_n_t: Yeah, but adding that conflicts with your own earlier statement of preserving the old behaviour "as much as we cold" (presumably, could)
07:18:55LloreanjhMikeS: My point is that adapting the current WPS to the new code should be a different discussion than "changing the behaviour of the default WPS"
07:19:12LloreanHe's basically saying he can't adapt it to the new code because he's unwilling to have it work the way it's always worked.
07:19:35S_a_i_n_tI'm not saying that at all...don;t be an ass.
07:19:36LloreanAnd he obviously doesn't have to adapt it.
07:19:44S_a_i_n_tI'm stating an opinion same as you are.
07:20:01LloreanS_a_i_n_t: ", I wish to the Gods on high that one way would "just work" so I can get on and start making the patch up."
07:20:05LloreanOne way *does* just work
07:20:12LloreanSo you *can* get on and start making it up.
07:20:40S_a_i_n_tNo, I can't, as we have no syntax.
07:20:51LloreanDidn't JdGordon_ give you the syntax for doing it that way?
07:20:59LloreanNest a pv progress bar inside a ?pv conditional
07:21:18JdGordon_that doesnt work in svn
07:21:20LloreanHe said the fix for it eating the nested tag stuff is minor.
07:21:26S_a_i_n_the gave several different versions of ways to do it
07:21:58S_a_i_n_tNothing is set in stone, which is why this is still a debate.
07:22:22LloreanOne requires what the JdGordon_ has said is a minor fix, then things can be made to work as they worked before only with more accuracy
07:22:27LloreanSeems like a pretty good starting point at least
07:22:45LloreanIt also is screen size and volume range independent.
07:23:04S_a_i_n_tyes, but untill I see actual code...there's no pint.
07:23:11S_a_i_n_tAs I may have to undo everything.
07:23:37 Quit liar (Ping timeout: 258 seconds)
07:24:05S_a_i_n_tHe may find another way of doing it, he may change his mind...I'm not writing code for something on a whim and a prayer.
07:24:31JdGordon_I mean not writing code on a whim and prayer is boring
07:24:59S_a_i_n_tbut, indeed, you *may* well change it completely...even you have to admit that ;P
07:25:16JdGordon_no, I think the pv inside pv is the way to go
07:25:40LloreanAt the very least it's a good halfway point until any other decisions are made.
07:25:46S_a_i_n_tand we're set on two images?
07:26:41*jhMikeS should probably actually learn WPS-eez and could contribute more meaningfully to all this stuff, since UI is always a contentious point, thereby providing a good fill of entertainment. :P
07:27:26JdGordon_and headache
07:27:47JdGordon_How does allowing , and ' for internal seperators sound?
07:27:59*jhMikeS has lots of naproxen sodium and good kidneys and liver to process it.
07:28:42S_a_i_n_t*and* ' ?
07:28:55JdGordon_not together of course
07:28:59*pixelma would rather want cabbie to display volume in decibels while changing volume like the builtin status bar does (and was started in the Clip version but never continued)
07:30:02S_a_i_n_tso, do we have a mute image?
07:30:07S_a_i_n_tthat's a no right?
07:30:15Lloreanpixelma: That is quite nice, yes.
07:30:26jhMikeSS_a_i_n_t: not that i know's exceedingly rare if so. :)
07:30:32LloreanS_a_i_n_t: Status bar displays mute, but I don't know if any of the cabbiev2 variants do
07:31:00jhMikeSthere's only one mute level anyway...mute
07:31:01S_a_i_n_tjust checking it wasn't going to be added with this.
07:31:21S_a_i_n_tsee? the original syntax I saw had mute.
07:31:36pixelmaS_a_i_n_t: it could be that monochrome or greyscale targets do, not sure
07:32:01S_a_i_n_tThat's why I'd like to see the syntax, seperators.
07:32:12S_a_i_n_tpixelma: IIRC, they won't anymore
07:32:21S_a_i_n_tbut I'm only working with colour for now.
07:32:21*jhMikeS wishes there could be a target with lower-than-mute volume to argue about, one with imaginary volume
07:32:23pixelmanot a "special" one, just because you can't represent "empty" in the same graphic style
07:32:43pixelmaS_a_i_n_t: what won't work anymore?
07:33:00S_a_i_n_twell, it will just be "empty" for mute.
07:33:06S_a_i_n_tunless I'm mistaken.
07:33:16pixelmawhy that?
07:33:30S_a_i_n_tswitch to progressbar style volume
07:33:52pixelmathe first part of %?pv is mute
07:34:06S_a_i_n_tand AFAIK, we're only doing images for min to 0dB and >0dB
07:34:17LloreanAnd min db should be 1 above mute.
07:34:22LloreanMute shouldn't show anything within the volume space.
07:34:34LloreanSince it's not really a "db value" at all anyway
07:34:53S_a_i_n_tpixelma: that's the *old* (yet still current) way it works.
07:35:10jhMikeSLlorean: a big "X" maybe?
07:35:19pixelmayes, so why can't the "new" way?
07:35:25LloreanjhMikeS: I'm happy with the current solution of "just show nothing"
07:35:34LloreanThere's no yellow at all, just the gray shadow behind the volume triangle
07:37:19S_a_i_n_tJdGordon_: so, it's " %?pv<%pv,volumebar.bmp,-,-,-,-|%xda> " ?
07:37:21jhMikeSLlorean: I know, but somehow, I don't feel quite perfectly certain by looking at it. I think, from an ignorant stance, "maybe it could just be quite near it since it's not accurate".
07:37:28S_a_i_n_tis that agreeable for now?
07:37:53JdGordon_pixelma: what do you tinhk about allowing the seperator for tags inside conditionals to be something other than | (like the example S_a_i_n_t just said where , is used instead of | )
07:37:55LloreanjhMikeS: Well, something more explicit isn't bad.
07:38:18LloreanjhMikeS: Especially since I think some targets still don't mute, and seeing a mute image but hearing very faint audio can let someone know to report that it's not muting the DAC yet.
07:38:51JdGordon_S_a_i_n_t: no, %pv doesnt change, so you still need the mute and 0 params, even if they are empty (which they shouldnt be or it wont work properly
07:38:52jhMikeSLlorean: perhaps show a little icon when it hits mute, otherwide hidden. For cabbie, it could go in that otherwise blank upper part.
07:39:35LloreanjhMikeS: Makes sense
07:39:41 Join JohannesSM64 [0] (
07:39:50S_a_i_n_tso, what are the params to be for mute anu 0? I thought we were using 2 images?
07:40:11pixelmaJdGordon_: I'm usually opposed to new character that need special treatment then when you want to use them "normally" (if a %, would be needed) and I don't see why it is here yet
07:40:23JdGordon_mute is either a image showing mute, or empty so the backdrop silooutte is used
07:40:44S_a_i_n_tand 0 needs its own image?
07:41:06pixelmaline level might be interesting
07:41:19LloreanAlmost definitely unless we know for sure that -1 will never show what looks like a full bar.
07:41:21JdGordon_pixelma: %?pv<xdaA|%pv,volume.bmp,-,-,-,-,|%xdaB|%xdaC> <- readable %?pv<mute|%pv|volume.bmp|-|-|-|-||0db|dander> <- unreadable
07:42:38S_a_i_n_tso, compared to volume.bmp, what is 0db's image to be?
07:42:50S_a_i_n_thow do we show that as an increment?
07:43:08pixelmaS_a_i_n_t: by the way - we have a mute button on most targets with a radio...
07:43:08S_a_i_n_tI'm *fucking* lost.
07:43:08JdGordon_it would be a copy probably of the full image
07:43:44S_a_i_n_tHmmmm, ok.
07:44:28 Quit linuxstb (Read error: Operation timed out)
07:44:33S_a_i_n_tSo, I'll also need to add a mute image, of should I just leave that out (for now)?
07:44:34JdGordon_I'm just testing it now
07:44:40LloreanJdGordon_: How would you distinguish it from -1 then?
07:44:42JdGordon_doesnt look like it works without a bmp
07:44:54S_a_i_n_tI was just thinking that.
07:44:58JdGordon_Llorean: you wouldnt, and you wouldnt be expected to
07:45:03S_a_i_n_t0 and -1 will be the same image.
07:45:12JdGordon_just like you cant tell the difference between -43 and -35
07:45:22S_a_i_n_tHmmm, fine with me. All i needed to know.
07:45:25 Join LinusN [0] (~linus@rockbox/developer/LinusN)
07:45:34S_a_i_n_t"do people expect it to be different."
07:45:39LloreanJdGordon_: So what, you can't tell when line level is hit?
07:45:52LloreanRight now 0db is easily distinguished from -1
07:45:58S_a_i_n_tyes, back off one from "fullred"
07:46:07LloreanS_a_i_n_t: Which only works on targets that have that.
07:46:14S_a_i_n_tgood point.
07:46:17JdGordon_so make the "full" not quite full
07:46:22LloreanAnd also makes it impossible to find without first going over, something you might not want to do.
07:46:27LloreanJdGordon_: Yes, that was my point actually
07:46:29S_a_i_n_tsee whay I haven't written code yet?
07:46:32 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
07:46:44Llorean-1 should stop 1 pixel or whatever visible unit short of the 0db image
07:46:56LloreanWhich just means you draw your progress bar 1 pixel shorter than the 0db image
07:47:09S_a_i_n_tthen we need three images, and just can't reuse one.
07:47:13LloreanOr 1 pixel short from the edge of the frame or whatever.
07:47:19S_a_i_n_tAnd the savins start becomming less.
07:47:42JdGordon_my laptop doesnt have internet making this hard....
07:47:43S_a_i_n_tipx is a LOT on some targets
07:47:49S_a_i_n_tand *nothing* on others
07:48:05LloreanS_a_i_n_t: I don't see how you can't reuse an image.
07:48:18LloreanYou could just draw the 1px short image + a single 1px wide image for the "0db" case.
07:48:19pixelmaJdGordon_: about your %pv explanation compared to %pb - with the progressbar you can just use %pb which will draw it in the current line, looks like this is not possible with %pv because then it is a number
07:48:33S_a_i_n_twell, it was sugested to reuse volume.bmp for 0dB
07:48:39LloreanI just said you could...
07:48:46LloreanYou reuse it plus append a single tiny image.
07:48:58JdGordon_pixelma: %pv|-|-|-|-| will do a boring line (you might need the height one filled in though
07:49:01LloreanAs in, you append it within the WPS rather than on the file
07:49:37S_a_i_n_tI am officially massively confused.
07:49:40S_a_i_n_tI have no idea what is happening anymore.
07:49:52S_a_i_n_tyou've all lost me.
07:50:43S_a_i_n_tWe're going to end up with four images including mute, and on most targets the savings wont weigh out with the hassle.
07:50:44jhMikeSthat's our real MO, to confuse and obfuscate /joke
07:51:19pixelmaJdGordon_: yes, but you compared the two as if the case %pb wouldn't exist (and using "the same code" I just want to remind of that just in case there could be problems in the code because you didn't think of this part)
07:52:09pixelmaand I still have my doubts about using the same tag name
07:52:31JdGordon_you are very kind, but no, you are thinking about it wrongly. %pb is the standard bar display, %pv is the standard volume display (i.e a number)... if you add extra params you get the new shiy display for both
07:52:33LloreanS_a_i_n_t: The only difference in size between the four images and the three images (without the 0db image) would be the size of the bitmap header.
07:53:02S_a_i_n_tso, what's the point?
07:53:16S_a_i_n_twhen it was one image I saw I'm not so sure anymore.
07:53:17JdGordon_the point is not needing 8 images for the variable bit
07:53:30 Quit anewuser (Quit: Like renoise + like music? 3 days to submit your entry!)
07:53:32LloreanS_a_i_n_t: The point is that it's also much smoother in increasing/decreasing
07:53:59S_a_i_n_tuntil 0db that is.
07:54:27LloreanYou have a mute image, a volume bar that's X pixels long, a 1px wide "0db" image that you draw alongside the full volume bard, and a "overboost" image.
07:54:30JdGordon_yes, so now you only need a mute, full, 0, clipping image, instead of 8 for full
07:54:49jhMikeSwhat about just switching to an alternate image and continuing to fill the bar as normal?
07:55:15LloreanInstead of 8 images all 40 pixels wide, you have 1 image 39 pixels wide, one 1pixel, and one 40 (if you have a full sized overboost image)
07:55:16JdGordon_or doing that so it is only 1 miage
07:55:20LloreanSo 1/4 the image data
07:55:23S_a_i_n_tjhMikeS: that was my "two image idea"
07:55:28S_a_i_n_tit got shot down
07:57:00 Quit Spaceghost (Read error: Connection reset by peer)
07:57:04S_a_i_n_tllorean, take a look at some of the smaller volumebars though
07:57:18S_a_i_n_tthat 1px is goint to be a LOT more than 1dB
07:57:35S_a_i_n_tthats why it doesn't work.
07:57:37LloreanjhMikeS: If we could take advantage of the cover art scaling code, you could possible just give it two arbitrary bitmaps and then tell it what size space for it to fit them in, and have it handle the transition through 0db depending on space available and range of the player
07:57:48JdGordon_on those I think going the multi colour single bar makes more sense and just ignore it not being exact
07:57:49LloreanSo the strip would always be 100px long, sometimes including a red bit sometimes not.
07:58:16JdGordon_and make it show the number when the volume is changing
07:58:17S_a_i_n_tJdGordon_: then you have no concept of line level
07:58:22jhMikeSS_a_i_n_t: Hmm...I've been dipping in and out. I'm not sure why it would be.
07:58:24S_a_i_n_tand people WILL cry about it.
07:58:44LloreanS_a_i_n_t: So draw an "Endcap" of some sort rather than making it appear to be "progress"... then it doesn't need to seem to be 1db and confuse people
07:59:25jhMikeSLlorean: Switching to an overdrive image though, allow ones without overdrive to display a full normal image, while in overdrive range, the switchover allows a full range without saturating it at 0.
07:59:39S_a_i_n_tThis is far more complicated than it needs to be, introduces special cases for targets...I'm losing faith.
07:59:45LloreanjhMikeS: I'm not sure I understood that statement.
07:59:49JdGordon_hmm... it looks like even with a bmp it draws one pixel at the minimum value
08:00:09LloreanJdGordon_: It doesn't when muted on my player at least
08:00:16LloreanBut mute+1 it shows a pixel
08:00:29JdGordon_the new code I'm talking about
08:00:38JdGordon_using the progressbar from cabbie to test
08:00:41*jhMikeS is cooking something and will return to explain shortly
08:00:46S_a_i_n_tJdGordon_: you're correct.
08:00:46LloreanS_a_i_n_t: I don't see why there's a problem with 1 pixel representing the last step to full.
08:00:58LloreanArguably the bar shouldn't "round up" to a 100% full bar until you're full
08:01:00S_a_i_n_tI had to hide it with an overlay in my OF theme
08:01:27S_a_i_n_tLlorean: well, what if the volume is 30px wide?
08:01:29JdGordon_or start the actual image 2 pixels into the bmp
08:01:30S_a_i_n_tor 11?
08:01:35S_a_i_n_tthen its a problem
08:01:41LloreanS_a_i_n_t: So you're saying it should present a full bar even at -10?
08:01:42S_a_i_n_tits either too much or too little
08:02:02LloreanIt's always going to have to approximate if there's not room. I'm just saying only the last step should "fill" just like it does on-target now
08:02:11LloreanThe bar isn't full until the 0db step
08:02:16S_a_i_n_tWe either need it to be consistent (like the current code) for targets...or, it's juat making things crazy.
08:02:58S_a_i_n_tAnd always showing 1px if its muted will be really offputting fot the "small volume=bmp" targets.
08:03:05Lloreanthe current code stops short of full, then only uses the full image for the single step of 0db. -1 is a different image, and +1 is a different image, independently of how many steps the -1 image covers.
08:03:19JdGordon_~line 1229 of apps/gui/skin_engine/skin_parser,c, can someone change that to "return ptr+1-wps_bufptr;" and test the bars inside conditionals please?
08:03:25JdGordon_that will stop it eating the whole line
08:03:26S_a_i_n_tyes, I do know how it works currently.
08:03:40LloreanSo why were you saying it needs to be "consistent (like the current code)"?
08:03:52LloreanMy suggestion was to mimic the current code's behaviour by only showing a completely full bar for 0db
08:03:58S_a_i_n_twhat I mean is, why introduce this this way and have special cases for some targets "just because it doesn't work"
08:04:09LloreanSince you can't draw a half pixel, you mimic this by only drawing the last pixel when the bar is full...
08:04:17LloreanWhat targets would be special cased?
08:04:24S_a_i_n_tIf it were just one image, it would be *so* much easier
08:04:31S_a_i_n_tbut it can't
08:04:41LloreanWhat targets would be special cased though?
08:04:47S_a_i_n_tthe "small volume.bmp" targets.
08:04:47jhMikeSLlorean: For mute, switch to mute image, for normal range, display normal image clipped according to full volume scale, above 0, display overdrive image clipped according to full volume scale.
08:04:54LloreanS_a_i_n_t: Why would they be special cased?
08:04:59JdGordon_how about.... %?mv<%pv|%pv|volume.bmp|....|> where that volume.bmp is a coloured bar
08:05:00LloreanDo they have less than 2 pixels for the volume image?
08:05:24S_a_i_n_tyou'd need to just vauely colourise the bitmap, instead of this stepping, and you'd loose all concept for libe level
08:05:41LloreanS_a_i_n_t: Is that what we do on those targets currently?
08:05:54LloreanI don't see why they'd need special casing for a smooth stepping bar when they don't need it for the coarse stepping bar.
08:06:07S_a_i_n_tno, but the way we want to do it now won;t work right.
08:06:23S_a_i_n_t-1px for less than line level is far too much
08:06:32S_a_i_n_tand on the bigger targets, far too little
08:06:34LloreanBut we currently do -1px for less than line level
08:06:34 Quit linuxstb (Quit: Leaving)
08:07:00LloreanAnd, the target would need to have more pixels for the image than volume steps before it becomes "too little" let alone "far too little"
08:07:27jhMikeSLlorean: so, for a target with no overdive, no OD image is shown and the normal image is shown fully at 0dB, while for overdrive targets, the image is shown gradually even in overdrive levels.
08:07:57S_a_i_n_tif the image only is 11px across, and you take away 1px for -1line's really about 3~4db-ish
08:08:09S_a_i_n_tprobably 2~3ish, but still
08:08:12LloreanjhMikeS: Maybe just draw the OD image on top of the full bar, as a sort of second bar? Just re-fill the progress bar to represent that it's "over full"
08:08:21LloreanSo you can then see when you're at 100% overdrive when the bar fills a second time
08:08:45LloreanS_a_i_n_t: If you have 8 steps and the image is 11 px wide, 1px = 8db.
08:08:58LloreanAnd yet we already use 1px as the change from -1 to 0
08:09:02LloreanOr sometimes several px
08:09:17LloreanThe problem you're stating we'd have is a "problem" we already have, and not really a problem
08:09:22jhMikeSLlorean: LOL. What? "Fills a second time"? I want to fill one time, no more than that! :)
08:09:30S_a_i_n_twhich is why if accuracy is one of the reasons this is's a fail.
08:09:52LloreanS_a_i_n_t: Displaying a 100% full bar when it's not 100% full is also inaccurate.
08:10:04pixelmaJdGordon_: (a) I'd rather see a [] or something around those multi-parameter tags (same with %St etc.) which would make the line more readable too (b) if you have need another separator, maybe you could reuse ; which isn't allowed in conditionals anyways IIRC - but I don't like introducing another separator for conditional because it seems inconsistent
08:10:05jhMikeSthat's sort of like puking during a binge so you can keep drinking ??
08:10:27LloreanS_a_i_n_t: 1px will never = 1db, they're always going to equal a range of them. So the single px below 0db represents -9 to -1 for example
08:10:30LloreanWhat's wrong with this?
08:10:37LloreanWhy should the last pixel have the same range as previous ones?
08:10:45pixelmaJdGordon_: (a) for every tag it applies to would break a lot of existing themes unfortunately I think
08:11:06S_a_i_n_tit shouldn;t, unless its a progressbar that is supposed to represent a percentile.
08:11:07LloreanjhMikeS: I'm just trying to think of a way to visually represent it without changing how much space you use.
08:11:20S_a_i_n_tin the way it is now, its fine.
08:11:26jhMikeSLlorean: so was I. it would do just exactly that with some minor code support.
08:11:37LloreanjhMikeS: I guess I don't well understand yours, sorry.
08:12:20S_a_i_n_tI don't understand it at all soon as it went past 2 images (or 3 for mute also) it kinda lost the whole point of it.
08:12:31LloreanS_a_i_n_t: Okay, think of it this way, "no pixels" obviously means mute. Once you've added X db, the first pixel lights up. Add X more the second one does. Keep adding X and eventually the last one lights up, and you're full."
08:12:49S_a_i_n_tLlorean: that is one problem already
08:12:52jhMikeSLlorean: the over-zero bitmap would be the same size as the one used for 0db and less, but *both* clipped according to the range, say, -89 to +6 dB.
08:12:53S_a_i_n_twith this way
08:12:55LloreanWhy does it need to keep going past lighting the last one up?
08:13:01S_a_i_n_twe don;t *have* no pixels
08:13:05S_a_i_n_tit's always 1
08:13:08LloreanS_a_i_n_t: JdGordon_'s fixing that
08:13:11LloreanOr did you miss that?
08:13:25LloreanOr at least, I think he is
08:13:29LloreanHe made it sound like it's unintentional
08:13:35S_a_i_n_tApparently I did...this is so fucking hard to follow.
08:13:52S_a_i_n_tSomeone needs to make up their mind, and even if I don;t like it, I'll do it.
08:14:05S_a_i_n_tbut we need to have our minds made up.
08:14:19LloreanjhMikeS: So on targets that go to 0db it'd fill the whole thing because the overboost bitmap would always be hidden, or?
08:14:26jhMikeSLlorean: yes
08:15:19jhMikeSthere, loading it could be avoided entirely with a quick check
08:15:35pixelmajhMikeS: not sure where you wanted to go there but not all audio settings are 0...n - bass and treble isn't
08:16:20S_a_i_n_tBasically, I'm at a point where I couldnlt give a rats ass how its done, as long as I know how it will be done, and that noone will change their mind.
08:17:12S_a_i_n_tBut then, someone will come along that wasn't in this discussion and say, why have you done this (for some reason I'm thinking kugel might not like it so much) and demand it be changed.
08:18:32S_a_i_n_tthe point it is now in SVN is fine if we just arbitrarily colour the bitmap for line-level
08:18:39S_a_i_n_tand it only usus 1 bitmap
08:19:03LloreanAnd you need target-specific bitmaps.
08:19:10LloreanRather than screen-specific
08:20:00S_a_i_n_tthat is less hassle than it is at the moment though
08:20:08S_a_i_n_tand more saving in buffer
08:20:26jhMikeSpixelma: are you referring to my earlier comment? The PP e200 has odd steps that the number doesn't reflect accurately, but it would be accurate and could just show whole numbers.
08:21:19S_a_i_n_tI'm honestly scared for the lefe of this project if it goes much past SVN now.
08:21:48S_a_i_n_tThe initial idea was to use 1 bitmap.
08:22:17jhMikeSLlorean: btw, you'd only need 3 bitmaps, mute, normal and, if supported "overboost". if we cut mute out, then 2 of course.
08:22:31S_a_i_n_tAnd in that sense, just arbitrarily colouring the bitmap to represent line-level (as we'll have to do for some targets anyway) seems to be the way to go.
08:23:22S_a_i_n_tnot entirely arbitrary of course, but it'd be just as accurate as we can get it if we *do *implement the curent way of thinking.
08:23:44LloreanjhMikeS: If the last step in the first bitmap was always the 1 step to line level, that should be fine.
08:25:33S_a_i_n_tDoes anyone have a nano1g (current SVN) sim?
08:25:46S_a_i_n_tI can show you how my idea looks, I'm already using it.
08:26:05jhMikeSLlorean: at least that could allow a completely accurate (if shown with enough pixels width) progress-style, working for any range
08:26:20S_a_i_n_tUsing only 1 bitmap really isn't as bad as it *seems* to be.
08:27:04LloreanjhMikeS: I'm really only concerned with "completely accurate" if "width in pixels" == "number of volume steps"
08:27:25LloreanIn any other width, I don't mind if there's a little inaccuracy introduced by being explicit where 0db is.
08:27:57LloreanThe first step past 0db should always present some sort of user feedback that something has changed.
08:27:58jhMikeSLlorean: exactly. it doesn't inhibit that if the themer wants it.
08:28:33LloreanjhMikeS: I guess I'm just not understanding the verbalization of what you're proposing. Somehow I can't imagine how it works in my head, sorry. I'll take your word for it though.
08:29:19*jhMikeS gets out the paint program real quick and tries.
08:30:54 Quit arbingordon (Quit: `)
08:31:27 Join ender` [0] (
08:33:06S_a_i_n_tthis is what I'm using in current SVN, works fine for me.
08:33:42S_a_i_n_tthe linelevel range is a *little* bigger than it need be, but otherwise fine.
08:34:33LloreanYou do realize line level is one *specific* value, not a range, right?
08:35:18 Quit TheSeven (Ping timeout: 258 seconds)
08:35:35S_a_i_n_tyes, I do, I mean the range on the bar, basically, if you're in orange you're close enough.
08:36:01S_a_i_n_tIt's going to have to be this way for some targets with the *new* implementation also.
08:36:11LloreanNo, it doesn't if things get fixed.
08:36:52S_a_i_n_tit was Jd's suggestion of "fixed" for "small volumebar targets"
08:36:56LloreanjhMikeS: So just have it transition to displaying the red bar on top at 1db?
08:37:12LloreanjhMikeS: I was trying to overcomplicated it in my head, I think
08:37:16jhMikeSLlorean: yeah, 1db transition, not 0db
08:37:33LloreanjhMikeS: Yeah, but it's still explicit about where 0db is, which is the most important part.
08:37:41LloreanAnd that can even let you know where 0db is inter-pixel
08:37:48LloreanEr intra-pixel?
08:37:55LloreanMan, I never know which of those prefixes to use in a case like this
08:38:19JdGordon_pixelma: well, i was tihnking the seperator could be either | or , or whatever so it is up to the themer and wont break current themes
08:38:27JdGordon_but yes using [] would be nice also
08:38:32LloreanS_a_i_n_t: If the progress bar ran from mute+1 to 0db, with the filled step being 0db, it wouldn't then be a problem on any size screen
08:38:38jhMikeSLlorean: Now you want it explicit ahead of time, not just when the user goes over?
08:38:44*JdGordon_ wouldnt be against breaking things there, or at least adding the option for them
08:38:49LloreanIt would work exactly substituting the multiple images we have currently in the conditional between mute and 0db
08:39:11 Quit BHSPitMonkey (Remote host closed the connection)
08:39:18LloreanjhMikeS: No, no. I said earlier I'd prefer explicit ahead of time, but the bare minimum is being able to identify it explicitly either way, which is what yours does. As in, that's more or less acceptable to me.
08:39:28*jhMikeS thinks he misread "explicit" and "isn't explicit" :\
08:40:03LloreanYours has the advantage of having the same scale left and right of the percentage, with an explicit 0db-1db changeover *even* if the progress bar is too small to see visible movement at that point.
08:40:11LloreanEr, left and right of the line mark
08:41:01CIA-5New commit by Buschel (r25601): Add a subsection for Replaygain to the runtime optimisation section.
08:41:04jhMikeSLlorean: ok, ok. gotcha. of course, I didn't explore all the artistic possibilities, just the basic idea. themers might do unexpected things.
08:41:11LloreanYeah, obviously
08:41:15LloreanThey always do.
08:41:27LloreanBut the ability to see the changeover independent of scale is an awful nice benefit to doing it that way, I think
08:42:06 Quit pixelma (Disconnected by services)
08:42:07 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
08:42:10 Quit amiconn (Disconnected by services)
08:42:12 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
08:42:27 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
08:42:34 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
08:44:01saratogaBuschel: replaygain is one multiply per sample, can that really have a measurable impact on battery life?
08:46:42 Join petur [0] (~petur@rockbox/developer/petur)
08:50:46S_a_i_n_tWhy don't we just couple 1 bmp with %?mv<||||%pv||>? that just draws on top of the bitmap so we know when it's linelevel.
08:51:03S_a_i_n_tseems like a WHOLE lot of hassle out the window from my point of view.
08:51:39S_a_i_n_t*I think there is one too many |'s before the %pv.
08:51:47S_a_i_n_tbut you know what my idea is.
08:52:34S_a_i_n_tyou could even just do%?mv<%pv|>
08:52:48S_a_i_n_tand know what every decibel change was.
08:53:04LloreanS_a_i_n_t: Pixelma brought that up quite a while ago
08:53:15S_a_i_n_tthen the .bmp doesn;t have to do any odf the work we need it to do, and complicates it by.
08:53:31S_a_i_n_twell, I must have missed it in the barrage of ideas.
08:53:37S_a_i_n_tseems the winner to me.
08:54:16LloreanIt still seems like there needs to be a way for the new feature to present the same information visually as the old one.
08:54:27LloreanjhMikeS' suggestion seems to do that quite well
08:55:21S_a_i_n_tactually (now that I recall) I believe she said to switch to numeric whilst changing volume, I'd rather it were drawn on top of the bitmap if it had to ba anywhere, and if you didn't want it, you could just leave the %?mv case blank.
08:56:30S_a_i_n_tthen, you'd always have a graphic/numeric volume, and always know if exceeding line level.
09:03:19 Join flydutch [0] (
09:05:00 Join clauwn [0] (
09:05:13clauwnhey there, how do i tell my rockbox to enter multimediamode`
09:05:34clauwni've got an ipod mini 2nd gen and it won't connect anymore
09:05:36S_a_i_n_twhat do you mean by multimediamode?
09:05:45S_a_i_n_tyou actually mean the HID?
09:06:15***Saving seen data "./dancer.seen"
09:06:15S_a_i_n_twell, that's not "multimediamode"
09:06:27saratogai think it is now
09:06:28clauwnit says starting multimediamode if i connect it
09:06:29saratogaor was
09:06:33S_a_i_n_tand it should always enumerate as a USB mass storage device
09:06:38clauwnbut it does not
09:06:48clauwndo you think i should restart my pc?
09:06:53S_a_i_n_twhat is the device?
09:06:56clauwni just started to copy a lot of files
09:06:58clauwnand aborted
09:07:03clauwnand now it won't reconnect
09:07:23clauwndevice is ipod mini 2nd gen
09:08:59S_a_i_n_tRebooting *may* help...I guess. In future, multimedia/mouse/browser/presentation are just different ways the player can enumerate to the host as an HID device.
09:09:12S_a_i_n_tAll of those modes support mass storage IIRC
09:09:48S_a_i_n_tMultimedia mode just lets you control your PC's volume
09:10:10S_a_i_n_tSkip tracks/fast forward/rewind etc.
09:10:19S_a_i_n_tMouse Mode is pretty obvious.
09:10:35clauwnok, i just rebooted my pc
09:10:39clauwnnow it connects again..
09:11:18clauwnand as i clicked it, it disappeared...
09:11:43clauwnerror while unmounting is : device is not a valid block device
09:12:00clauwnoh no, while mounting
09:12:08S_a_i_n_tHmmm. that's new to me.
09:12:25clauwni'll probably reinstall rockbox later :D
09:12:30S_a_i_n_tYou may need to wait to see if anyone else has any ideas about this.
09:13:06clauwnmight be a problem with my pc, as the entry for apple music-player won't disappear if i disconnect it
09:13:30clauwnthank you
09:13:33clauwni'll come back later
09:13:56pixelmaJdGordon_: I understood the either | or , but find that confusing and inconsistent
09:14:22clauwnso i'm off
09:14:26*clauwn detaches
09:14:46 Join wodz [0] (
09:15:38pixelmaclauwn: you could try turning HID off
09:19:53JdGordon_pixelma: yes inconsistant, but MUCH more readable
09:22:46 Join lpereira [0] (
09:23:00pixelmayou could have had a chance to introduce [] or similar for this volume bar tag if you had given it a bit more time and thought in the beginning
09:23:35pixelmaand as I said ; might work there too and already is a special character and not allowed in conditionals otherwise
09:25:35JdGordon_; wont make things easier to read
09:25:53JdGordon_and really, bitching about speed isnt going to change anything
09:25:58amiconnJdGordon_, Llorean: Regarding the volume bar above 0dB, couldn't the bar just use two bitmaps, each representing the full range, one in e.g. green and the other in e.g. red, and then the token switches bitmaps at 0dB, but regardless of which one is used, draws the bar as normal?
09:26:13S_a_i_n_tNothing short of embedded TTS wil make this line easier to read ;)
09:26:23S_a_i_n_tBut, it could certainly be worse ;)
09:26:30amiconnThe only disadvantage I see with this approach is that targets which can't go above 0dB (e.g. the irivers) have a useless bitmap
09:27:07S_a_i_n_tbut, they stilll save many a bitmap
09:27:13S_a_i_n_t~7 or so.
09:27:30S_a_i_n_t~bah...not 7, ~5
09:28:03jhMikeSamiconn: I was thinking the loader could skip accessing it by checking first if > 0dB is not available.
09:28:14JdGordon_haha /me puts %?pv<mute|%pv|volume.bmp|-|-|-|-||0db|dander> through OSX's TTS...
09:28:43S_a_i_n_tI bet thout sounded, messed up.
09:29:09*jhMikeS supposes that WPS memory is more valuable, byte per byte than a small bit of disk space
09:30:14JdGordon_amiconn: in all honesty, I dont see what the point of the whole disucssion is. We are talking about a graphical representation using maybe 1/10th of the pixels which would be needed for an accurate display
09:30:24JdGordon_IMO a single one with red ends are good enough
09:31:03S_a_i_n_tI think so too, but everyone cares about where linelevel is too much
09:31:09amiconnImo it does matter whether you're +1dB or +12dB above...
09:31:38S_a_i_n_tamiconn: JdGordon_: At last! Sanity!
09:31:39amiconnIf you want line level you just load a .cfg and don't change volume in the wps
09:31:43jhMikeSJdGordon_: unless the user wants a fully accurate bar. then at least all appetites are satisfied. if they don't want to use it, just use the same .bmp twice? It only loads a particular image once, correct?
09:31:53JdGordon_the only correct way to do this is showing the actual number when it is changing
09:32:21S_a_i_n_tJdGordon_: pixelma both had a go at suggestion that.
09:32:42S_a_i_n_t%pv, then %pv|playbar.bmp|
09:32:46JdGordon_jhMikeS: no, it isnt that smart, it will load the same bmp more han once
09:33:09JdGordon_but anyway, we are talking about the default WPS, all other ones will do whatever the themer wants
09:33:19jhMikeSJdGordon_: you can't load some bar bitmap, give it a label, and then use it for both normal and over, and only load it once?
09:33:40S_a_i_n_tpreferably, I'd perfer the numerical time to be drawn on top of the bitmap volumebar.
09:34:06S_a_i_n_tBut, as long as it's there...
09:34:28JdGordon_jhMikeS: that could be done, but the way the bar was done was with the smallest possible code change
09:34:30JdGordon_so no
09:34:38 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
09:35:03 Quit skx` (Ping timeout: 248 seconds)
09:35:27*JdGordon_ thinks maybe asking for cabbie to be updated wasnt such a good idea
09:36:31 Join Bagder [0] (~daniel@rockbox/developer/bagder)
09:36:50S_a_i_n_tIn hindsight S_a_i_n_t agrees
09:37:10wodzSomeone posted somewhere configuration line for vim to conform with rockbox coding style. Do You recall this line?
09:37:18S_a_i_n_tperhaps just leave the SVN volumebar as an option open to themers ;)
09:37:23jhMikeSLOL. Well, can't say it wasn't intense. :)
09:37:41S_a_i_n_tjhMikeS: that's one word for it.
09:37:53S_a_i_n_tI have a massive headache now :(
09:39:12*JdGordon_ wonders how hard it would be to get a script going to fix all the online themes if syntax is changed
09:40:52 Join skx` [0] (
09:45:22 Join stripwax [0] (
09:47:01 Quit stripwax (Client Quit)
09:47:36jhMikeSJdGordon_: the old way with an image list won't be supported?
09:47:56JdGordon_it is
09:48:00JdGordon_nothing was removed
09:48:31JdGordon_if you want a fluid volume bar it isnt so easy to find mute/0/>0
09:48:37JdGordon_that is the whole discussion
09:50:16jhMikeSit certainly was but I wondered if it wasn't supposed to replace other ways and be the "only way". :)
09:50:45JdGordon_no, no :)
09:54:42 Quit shai (Quit: Leaving)
09:58:51pixelma JdGordon_: bitching about speed? You mean the fact that you rushed this in? I just said that there was a chance to do things differently
09:59:24wodzamiconn: ping
09:59:48pixelmaand I don't see where the difference between ; and , would be in regards to readability of that like
10:04:11 Quit ender` (Ping timeout: 276 seconds)
10:09:03 Join shai [0] (
10:10:50 Part lpereira
10:11:58 Quit avn (Ping timeout: 276 seconds)
10:12:03JdGordon_is %xd and %Vd the only tags which can have another letter on them that doesnt use |'?
10:12:38JdGordon_%d %D also
10:13:21 Join avn [0] (~avn@
10:13:23S_a_i_n_twhat is &D?
10:13:31S_a_i_n_tI don;t remember that one...
10:13:33JdGordon_next tracks dir
10:14:20JdGordon_and the timeouts
10:15:49 Join ender` [0] (
10:16:23JdGordon_I really want to redo the parser, and dissallowing those sort of tags would make it much simpler
10:19:47 Join ender [0] (
10:20:44 Quit ender` (Ping timeout: 264 seconds)
10:21:33 Join kugel [0] (~kugel@rockbox/developer/kugel)
10:29:59 Quit avn (Ping timeout: 248 seconds)
10:31:56 Join avn [0] (~avn@
10:32:58 Quit shai (Ping timeout: 245 seconds)
10:34:55 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
10:36:55 Quit linuxstb (Client Quit)
10:40:35 Join shai [0] (
10:46:28 Join Schmogel [0] (
10:51:05kugelwouldn't %?pv<%pv|green-volume.bmp|-|-|-|-||%pv|orrange-volume.bmp|-|-|-|-||%pv|red-volume.bmp|-|-|-|-|> work?
10:51:16kugelw.r.t to tonights discussion
10:51:55Lloreankugel: You missed the mute part of the conditional, but yeah, it seems like it ought to.
10:52:03LloreanGreen being <0, orange being =0, red being >0
10:52:16kugelyes, that's what I thought
10:52:38LloreanAnd that accomplishes the effect with existing code (+ the %pv inside %?pv fix)
10:54:04kugelit would be even better, the +6db part would be rendered smoothly
10:55:30LloreanAnd still uses less bitmaps / bitmap data than the current cabbiev2 code
10:55:30kugelI think someone should try if it works :)
10:55:30 Part LinusN
10:55:36LloreanWell, I think it can't yet because I think you can't yet put a %pv inside of a %?pv
10:56:53kugelthe other part of the commit was to allow multiple progress/volume bars per viewport, right?
10:57:17kugelso if it doesn't work it's probably only a parser problem, not an actual real problem
11:00:33wodzadding keypad definition to plugins drive me crazy :-/
11:02:41Lloreankugel: Yeah, IIUC it's exactly a parser problem
11:03:12 Quit kugel (Ping timeout: 276 seconds)
11:03:20 Join bluebrother [0] (
11:03:20 Quit bluebrother (Changing host)
11:03:20 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
11:06:19***Saving seen data "./dancer.seen"
11:06:56 Quit bluebroth3r (Ping timeout: 260 seconds)
11:07:54 Join kugel [0] (~kugel@rockbox/developer/kugel)
11:08:24 Quit tipi^ (Ping timeout: 276 seconds)
11:08:27 Join tipi^ [0] (
11:10:32 Join LinusN [0] (~linus@rockbox/developer/LinusN)
11:11:53 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
11:12:42JdGordon_Llorean: kugel yes, its a one line fix, but my laptop doesnt have internet atm so cant commit it
11:14:37 Join Kitr88 [0] (
11:17:49 Quit Kitar|st (Ping timeout: 258 seconds)
11:21:07 Join pyro_maniac [0] (
11:21:36pyro_maniaccommiters around? i got a online fix for lang.make
11:23:33kugelI'm around yes
11:24:47pyro_maniacon my up-to-date debian tail decline "tail -1" as deprecated and has to be replaced by "tail -n 1"
11:28:34kugelwhy's that deprecated :(
11:29:42wodzit is for quite long time
11:29:57Bagdermy up-to-date debian tail supports it fine
11:30:08Bagderand yes, its has been said to be that for ages
11:31:53 Join efyx [0] (
11:33:12pyro_maniaci can only report what my system said:
11:34:31 Quit kugel (Ping timeout: 276 seconds)
11:36:20 Join webguest17 [0] (
11:39:47pyro_maniacthis is more complete:
11:40:07 Join Lixun [0] (~8984fa0d@gateway/web/freenode/x-hwhpjiqfmwahujsv)
11:40:07 Quit webguest17 (Client Quit)
11:42:50ZagorBagder: are you on testing or unstable?
11:43:01Bagderboth ;-)
11:43:16BagderI checked two testing and one unstable
11:43:24Zagortail −−version
11:43:42Bagdertail (GNU coreutils) 8.4
11:43:55Zagorok. pyro_maniac what does your version say?
11:44:08Bagdermy testing says 7.4
11:44:20Zagoryes. "trying to update coreutils from 7.4-2 to 8.4-2 (candidate is 36 days old)"
11:45:54pyro_maniachmm, i am confused. mine said "tail (GNU coreutils) 5.3.0"
11:46:49 Quit phanboyiv (Read error: Connection reset by peer)
11:47:37pyro_maniacbut i am using debian testing too
11:47:49Zagorpyro_maniac: dpkg -l coreutils
11:49:50pyro_maniacthanx for help and sorry for this not really rockbox related thing
11:51:09Bagderwe could still switch to -n 1
11:54:11 Quit avn (Ping timeout: 276 seconds)
11:55:36 Join avn [0] (~avn@
11:55:52 Quit Lixun (Quit: Page closed)
11:56:27TorneBagder: we should, really..
11:57:03 Join kugel [0] (~kugel@rockbox/developer/kugel)
12:16:40 Quit avn (Ping timeout: 252 seconds)
12:18:16 Join avn [0] (~avn@
12:20:17 Join m3dlg [0] (~m3dlg@
12:25:51CIA-5New commit by funman (r25602): fix yellow
12:26:15 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
12:27:19 Join funman [0] (~fun@rockbox/developer/funman)
12:27:57funmanI want to disable adjustable_cpu_freq on as3525v2, thoughts ?
12:31:12funmanTorne: can you point me again to that file in symbian source which lists the capabilities for various ARM cpus ?
12:31:24Tornefunman: yeah, one sec
12:31:28Tornewe don't use the test-and-clean operation
12:31:35Tornei didn't know it existed until i looked it up yesterday
12:31:42kugelI see no reason to disable it, but I think disabling it makes us lazy about fixing it properly
12:31:56funmankugel: it's too buggy
12:32:01funmanTorne: FS #11106
12:32:38funmankugel: having it enabled doesn't make us fix it faster
12:32:45Tornefunman: yeah, i had a look and it seems reasonable, but i don't know anyting about it really
12:32:49Tornefunman: i assume it's faster?
12:33:01funmanTorne: I don't know if it's faster, I just know it's correct
12:33:20Torneer, how is the current code not correct?
12:33:22funmanarm926ej-s use a different index format for cleaning/invalidating
12:34:26funmanhmm it's restricted :/
12:34:32Torneoh, er, sorry
12:34:33Tornewrong path
12:34:36Torneone mo
12:35:04Tornethere are two copies of the code, one is not world readable
12:35:08Tornebut i have access :)
12:35:40Tornetry that?
12:35:50funmanworks, thanks!
12:36:08Tornethe actual cache maintenance code may be interesting also
12:36:23Tornebut again, symbian is EPL so you can't actually lift the code
12:36:49Torne <- that's most of the asm for cache maintenance
12:37:15Tornewe clean by way-set index. and i assure you it works.. :)
12:37:46funmanfor arm922tdmi the index format is 31:26 = index, 6:5 = segment; for arm926ej-s it is 31:32-A = way, S+4:5 = Set (=index), 4:2 = Word
12:38:18funmanhm so it might just mean that set/way are variable
12:38:37Tornethey are known, though
12:38:51funmanbut the index has another position, so the 2 formats aren't compatible afaiu
12:39:05 Quit ender (Ping timeout: 258 seconds)
12:39:10Tornewell, we hav ejust one implementation that works on all ARMs, afaik :)
12:39:34Tornethough it's kinda hard to follow
12:39:56Tornethe cache code is slightly nightmarish because it's allowing for external cache, and SMP, and so on
12:40:17 Join ender` [0] (
12:40:22funmanI see simple asm one-liners after line 200
12:40:53Torneonly for ARM7 and StrongArm
12:41:23funmanline 242: elif defined(__CPU_ARM920T__) || defined(__CPU_ARM925T__) || defined(__CPU_ARM926J__)
12:41:33funmanno clean/invalidate by index though
12:42:04Torneer, it also uses weird terms for the operations
12:42:13funmanthese use MVA
12:42:15Torneflush is clean and invalidate
12:42:23Tornepurge is invalidate only
12:42:28Torneso FLUSH_DCACHE_INDEX
12:42:30Torneis the one
12:42:34Torne(i think)
12:42:40funmanoh right
12:42:55funmanmissed it because it hadn't the same alignement =)
12:43:17Tornebut yeah, the code that uses these macros is in the other file i linked
12:43:26Torneit's parameterised by the cache line/index/etc sizes
12:43:30Tornewhich it gets from.. somewhere.
12:43:37Tornepossibly from cp15?
12:43:43Torneor mabe just provided by platform headers.
12:44:39funmancode/data at a fixed position always gets loaded in a determined cache line, right?
12:45:25 Quit JohannesSM64 (Ping timeout: 268 seconds)
12:45:39funmanif you can't know the cache line where some piece of code get loaded you'd use this index only in a loop to flush the whole cache then?
12:45:51TorneFor remappings/etc you do it by MVA
12:46:21TorneA given address in memory is mapped to a particular set index, but the data can be in any way of that set index
12:46:37funmancan you grep for (CLEAN|FLUSH)_DCACHE_INDEX and look for 2 different formats (ARM920T/ARM926J) ?
12:46:51Tornethere aren't two. i linked the code that does it.
12:46:59funmanhm ok
12:47:35Torneit is parameterised by a set mask, line length, cache size in lines, and "way increment value"
12:47:46Tornesee InternalCache::Clean_DCache_All() in cache_maintenance.cia
12:47:54funmanhm is that code to be used by external users?
12:48:02Tornedefine external
12:48:07funman3rd party drivers
12:48:07Tornethis is not exported from the kernel
12:48:15TorneDrivers get a vastly simplified interface
12:48:28Tornewhich abstracts away all details of how many layers of cache exist, etc
12:48:33 Join mitk [0] (~mitk@
12:48:40TorneInternalCache only does ARM L1 CP15-driven caches
12:48:46Tornebut most symbian targets have L2 cache now
12:49:08 Quit m3dlg (Ping timeout: 260 seconds)
12:49:18Torne has the exported interfaces for drivers
12:49:25Tornethey look like "SyncMemoryBeforeDmaRead" and the like :)
12:49:26funmani download the tip.bz2 to check
12:49:52Tornethe assumption is generally that drivers only do cache operations for DMA coherency, or possibly as IMB
12:50:03Tornethey don't get given low level operations like clean/invalidate/etc
12:50:17Tornethey are supposed to communicate intent only and the kernel does the right hting :)
12:51:29funmanthe code in cache_maintenance.cia depends on iCleanAndInvalidateMask and iLineLength
12:53:14funmanin SCacheInfo (there must be static definitions of this somewhere)
12:53:36Torneyeah, grep for it
12:53:41Tornei don't remember where it gets set up
12:54:06Torneer, no
12:54:12Tornethat's all L2 stuff
12:54:46Torneit's quite possibly set byt he baseport, which will be a problem since there isn't a decent published baseport there :(
12:54:59Torneall the baseports i use are IP-encumbered still
12:56:14funmancache_maintenance.cpp::InternalCache::Init1() ?
12:57:04Torneyeah, there you go ;)
12:57:32Torneso yeah, it's extracting it from cp15 cache info registers
12:57:40Tornevia a slightly hilarious process :)
12:58:45funmanhm so the format could be compatible after all
12:58:56 Join JohannesSM64 [0] (
12:59:18funman"For all the cache operations, the Word should be zero"
12:59:43mitkfunman: There is no rockbox-info.txt after compiling clip+ with test plugins. Is it intentional?
12:59:54funmanmitk: i don't think so
13:00:29funmanwhat make $PWD/rockbox-info.txt gives you ?
13:01:57mitkFunman: gives me rockbox-info.txt \o/
13:04:25 Quit ender` (Ping timeout: 246 seconds)
13:04:29funmantest plugins do not build here
13:06:22***Saving seen data "./dancer.seen"
13:07:10kugelsome need #ifdefing for certain targets, I didn't double check all
13:08:28mitkfunman: you mead it doesn't make sense to build it yet or there are errors during compilation? I can build it and I have 2 test plugins after compilation: test_boost and test_codec.
13:09:09CIA-5New commit by funman (r25603): some test plugins have dependencies
13:09:35kugeltest_resize is lcd_color?
13:09:37funmanmitk: what kugel said
13:09:56funmansmooth_* are lcd color (in plugins/lib/SOURCES)
13:09:59kugelgreyscale has albumart with scaling too, afaik
13:10:06 Join ender` [0] (
13:10:17kugelah, I thought it tests the other scalers as well
13:11:03funmanmitk: rockbox-info.txt was probably not built because of errors, it builds fine there (with last commit)
13:16:55Tornefunman: the test-and-clean operaation might still be better if the cpu supports it
13:17:01Tornefunman: if nothing else the code is way simpler
13:17:12Tornefunman: i don't know why symbian doesn't use it.
13:17:22Tornethere may be a reason, or we may just not have noticed :)
13:17:25funmanbut i was under the assumption that this would fix some problems with the Clipv2/Clip+
13:17:39TorneWell, possibly, if the current loop is not successfully cleaning all the sets
13:17:59Tornewhich it may not be, it has too many hardcoded constants to work on *all* ARMs. I don't know about the specific core in the clipv2
13:18:06funmanthe index format in arm922t reference manual is a bit misleading
13:18:13funmanTorne: well it's an arm926ej-s
13:18:14TorneThe processor reference manuals suck
13:18:22Tornethey are context-free :(
13:18:54Tornei am spoilt by being able to email and get answers, generally ;)
13:19:02 Join archivator [0] (~archivato@
13:19:33 Join slck [0] (Venci@Slackware.SlackPix.Com)
13:20:34 Join evilnick_ [0] (
13:21:21funmanrockbox mmu-arm.S says: 31:26 = index, 7:5 = segment; arm922tdmi ref manual says: 31:26 = index, 6:5=segment; arm926 ref manual says: something else (no mention of segments anymore)
13:21:31mitkfunman: I had no errors compiling with test_plugins till few commits back. But now it doesn't matter I think. Thanks.
13:22:27 Quit archivator (Client Quit)
13:24:11funman31:26 + 7:5 => 9 bits for the index, that would mean a 16kB cache with 32bits line size, or a 8kB cache with 16 bits line size
13:26:57 Join halmi [0] (~Miranda@
13:27:11funmanworks for the gigabeatf
13:27:56kugeli think the caches have 32*bytes* like size, don't they?
13:28:15funmanthe PP have 16 bytes
13:28:15 Join luistmw [0] (
13:28:28kugelthat's non-standard arm cache though
13:28:50funmangigabeatf (arm920t) has 16kB caches, as3525 (arm922tdmi) has 8kB caches
13:28:57kugelf/x has 8k+8k cache IIRC
13:29:03luistmwhey can som1 help me
13:29:12 Quit ender` (Ping timeout: 245 seconds)
13:29:23luistmwi think i bricked my nano 2g
13:29:41luistmwi erased the .rockbox file
13:29:55luistmwwhat can i do
13:30:14funmankugel: where did you see that?
13:30:25luistmwin its main folder
13:30:33 Part luistmw
13:30:34kugelfunman: 920t reference manual
13:30:50 Join DataGhost [0] (
13:30:50 Quit DataGhost (Changing host)
13:30:50 Join DataGhost [0] (~dataghost@unaffiliated/dataghost)
13:30:52jhMikeSfunman: I think there's differences there. See DDI0151C_920T_TRM.pdf p.45 for index format
13:30:58funmankugel: says 16kB
13:31:14jhMikeS7:5 = seg
13:31:19kugelyes, split into 8K data and 8K ins cache
13:31:29funmanjhMikeS: true, 1 more bit is needed
13:31:40funmankugel: no 16kB dcache and 16kB icache
13:31:46kugelhm, looks like I'm wrong
13:32:01funmanjhMikeS: as3525 has only 8kB cache, twice less than gigabeatf/x
13:32:33Tornefunman: arm9 is almost guaranteed to be 32 byte lines
13:33:00funmanstill i don't see how arm926 and arm920/922 index formats are compatible
13:34:04kugel8K+8K on 922t then
13:35:00funmanDDI0198 page 52 (chapter 2-22) gives an example for 16kB 4-way set associative cache: 31:30 = way, 11:5 = index
13:35:38 Join ender` [0] (
13:35:57funmanhm but for 64-way it's the same format
13:37:12jhMikeSfunman: I suppose those routines just need half as many in the unrolled loop
13:37:20funmanarm920/arm922 use 64-way caches, but arm926 use 4-way
13:37:28funmanjhMikeS: yes, not critical though
13:38:02 Quit Zarggg (Ping timeout: 260 seconds)
13:38:03jhMikeSfunman: might it cause problems where it's not valid?
13:38:24funmandon't think so, the as3525(v1) work fine
13:38:50 Quit halmi (Quit: halmi)
13:39:05funmanwe should correct it anyway, perhaps set the cache size in cpu.h
13:39:35funmanTorne: are you aware of weird problems occuring with arm926ej-s phones using symbian ?
13:39:48Torneno :)
13:39:57TorneI am certain that our cache code works
13:40:07Torneat least on <=Cortex-A8
13:40:17Torneit may no longer work on ARM7, also
13:40:20Torneas that may have rotted ;)
13:40:53funmanas i understand it, your code (and ours) would not flush some sets
13:42:11funmanhum wait
13:42:12jhMikeSthe final sub instruction before the adds can be taken out of the loop there as well, then a constant defined
13:42:13Tornethe result of that would be catastrophic, though, no?
13:42:23Tornei am certain we would have noticed :)
13:42:34funmanTorne: you're right, you read the settings from cp15 so it's ok
13:42:53funmanjhMikeS: we could use a cache_init() function to read the cache properties
13:42:55 Join luistmw [0] (
13:42:55 Part luistmw
13:43:01 Join Zarggg [0] (
13:44:15funmanfrom c0, c0, 1
13:44:16jhMikeSfunman: we could, but then again, they're constant for any player and the routines can be optimized
13:45:05jhMikeSunless...I'm not aware of some sort of difficulty with telling at compile-time ??
13:46:00funmanthe difficulty is figuring out the values for each player
13:46:12funmancowond2, mini2440, mrobe500, and creative zen vision
13:47:17Torneer, surely that's not a difficulty
13:47:26jhMikeSyou just said how :) splash it up somewhere, then you've got it.
13:47:26Tornedump the cache id registers in the debug screens
13:47:39Tornethen shove it in a target header
13:47:56funmanwell it's probably easier to have 2 settings for as3525 and gigabeat, and use the other method on arm926ej-s
13:48:05 Join krabador [0] (~darkham@
13:49:30funmanjhMikeS: FS #11106 : any thought on adding the config option in configure ?
13:50:00 Nick fxb__ is now known as fxb (
13:51:14funmanclipv2: cp15, 0, %0, c0, c0, 1 == 0x1d152152
13:56:16funman> 16kB cache
13:57:41CIA-5New commit by funman (r25604): test_codec is SWCODEC only
14:00:10funmansame on clip+/fuzev2
14:00:17 Join TheSeven [0] (~theseven@rockbox/developer/TheSeven)
14:01:34funmanoops I was wrong, mini2440 uses the same CPU than gigabeat F/X
14:02:06 Join dfkt [0] (dfkt@unaffiliated/dfkt)
14:02:07CIA-5New commit by uchida (r25605): viewer plugin: when the alignment is RIGHT, supports WIDE screen.
14:02:12 Join liar [0] (
14:05:47jhMikeSfunman: certainly interesting, using mrc with pc. why's it that way?
14:06:29funmanthat's how it's documented in arm926 ref manual (pc itself isn't modified)
14:06:41TornejhMikeS: "magic"
14:07:03TornejhMikeS: it's because ARM used to have the flags in PC, back when the program counter was 26-bit
14:07:09jhMikeSI look at a later patch and it looks completely different.
14:07:30jhMikeSTorne: way before my time with ARM, I can say that much
14:07:47TornejhMikeS: yeah, this was ARMv3 and prior
14:08:20Tornethe use of pc to refer to the flags in weird special case operations was left over in various odd places, and this one has survived :)
14:09:42funmansurvived, or added for some nostalgia ? :)
14:10:56funmanoops i made too much copypasta for invalidate_dcache()
14:12:00jhMikeSfunman: for me, it's like saying heiroglyphs are nostaligic (perhaps in pastlife regression) :)
14:12:42 Quit avn (Ping timeout: 252 seconds)
14:13:35jhMikeSeverythings running slow and swapping here, so my responses might be delays (too much junk open)
14:14:25 Join avn [0] (~avn@
14:14:59 Join DataGhost_ [0] (
14:14:59 Quit DataGhost (Disconnected by services)
14:15:00 Nick DataGhost_ is now known as DataGhost (
14:15:37CIA-5New commit by uchida (r25606): fix yellow
14:17:10jhMikeSfunman: I think it looks good. I'd need to apply it to know for sure and will take your work for the unfamiliar stuff.
14:18:59 Join DataGhost_ [0] (
14:18:59 Quit DataGhost (Disconnected by services)
14:19:00 Nick DataGhost_ is now known as DataGhost (
14:20:50funmanjhMikeS: last patch on FS #11106, I just added a comment compared to the last paste
14:23:37funmannot sure why we didn't spot any problem with cowond2/mrobe500 and creative zen
14:24:03 Join Watermark [0] (
14:24:05funmanalso as3525v2 worked "pretty fine"
14:24:38 Quit Watermark (Client Quit)
14:25:57 Nick markun_ is now known as markun (~markun@rockbox/developer/markun)
14:29:29jhMikeSfunman: FWIW, a small optimization to the bulk routines which I think would conflict w/patch:
14:30:28 Quit mitk (Quit: Leaving)
14:30:36funmani thought about it, i'm ok with this
14:31:48funmanis it better to use mov+sub to calculate r1, rather than a ldr ?
14:34:06funmanbtw, =(0x4000000 - 7*(0x20)) is easier to understand IMO
14:35:28 Join nimak [0] (
14:36:20 Join halmi [0] (
14:36:50 Quit halmi (Client Quit)
14:37:18 Quit krabador (Read error: Connection reset by peer)
14:37:35jhMikeSfunman: I was thinking that too, about calc'ing the value with the preprocessor.
14:38:06 Quit nima (Ping timeout: 264 seconds)
14:38:33funmanyou can use multiplications/additions/substractions/shifts (probably divisions and modulo) with the assembler
14:38:37jhMikeSor whatever works, less #ifdef stuff. ldr on 920T can work wile r0 is zeroed out
14:39:30jhMikeSless #ifdef and one less insn in the loop = cleaner + faster (slightly), no?
14:40:11funmanhm how do you remove an #ifdef?
14:41:08funmanldr r1, =(0x4000000 - SOMETHING_DEFINED_IN_CPU_HEADER*0x20) ?
14:43:16jhMikeSyeah, something like that, for the number of segs
14:44:00 Nick YPSY is now known as Ypsy (
14:44:12 Join froggymana [0] (
14:44:24funmansaratoga: ping
14:46:41 Join _jhMikeS_ [0] (~jethead71@rockbox/developer/jhMikeS)
14:46:41 Quit jhMikeS (Disconnected by services)
14:46:45 Join M3DLG [0] (~M3DLG@
14:46:48_jhMikeS_ldr r1, =(0x04000000 - (SOMETHING_DEFINED_THAT_CAN_ALSO_SELECT_THE_ADDITIONAL_INSTRUCTIONS - 1) * 0x20)
14:47:04 Quit kugel (Ping timeout: 276 seconds)
14:47:10 Nick _jhMikeS_ is now known as jhMikeS (~jethead71@rockbox/developer/jhMikeS)
14:47:14funmansounds good
14:51:21 Part LinusN
14:51:59 Quit n17ikh (Ping timeout: 276 seconds)
14:52:10 Join MasterFen [0] (
14:52:34MasterFenHello there everyone!
14:53:47MasterFenI just installed Rockbox on my Clip+ and I'm willing to beta test if needed.
14:54:06 Part MasterFen
14:55:27 Join MasterFen [0] (
14:55:41 Part MasterFen
14:57:15 Join n17ikh [0] (
14:59:45 Quit avn (Ping timeout: 258 seconds)
15:01:29 Join avn [0] (~avn@
15:01:33jhMikeSfunman: I suppose now I'm going off the deep end. Maybe mov r1, #0x04000000 ; sub r1, r1, #((SOMETHING-1) * 0x20) would be faster, since it's prefetched already. lol. who knows, I've been awake all night.
15:02:07 Quit antil33t (Read error: Connection reset by peer)
15:02:13 Join antil33t [0] (
15:02:49funmanI think it's not critical anyway, don't lose too much neurons on that :)
15:04:26funmansaratoga: Did you edit because you thought testing wasn't needed anymore ? backlight now works but there still could exist Fuzev2 with the other LCD type, so only the backlight would work. (If it was the reason please remove your edit)
15:05:15jhMikeSfunman: It's a zen thing, I'm allowing loss of the ego in order to be aware of infinite possibilities and the emptiness of the code, that is all. :)
15:05:53 Nick Ypsy is now known as YPSY (
15:06:03 Quit M3DLG (Ping timeout: 252 seconds)
15:06:25***Saving seen data "./dancer.seen"
15:06:35 Nick YPSY is now known as Ypsy (
15:09:09 Nick Ypsy is now known as YPSY (
15:10:55 Nick YPSY is now known as Ypsy (
15:10:55wodzis it possible that not having any theme at all may introduce some performance penelty?
15:11:27 Nick Ypsy is now known as YPSY (
15:12:35 Nick YPSY is now known as Ypsy (
15:13:26 Join archivator [0] (~archivato@
15:16:29 Quit wodz (Quit: Leaving)
15:19:25 Quit avn (Ping timeout: 246 seconds)
15:21:06 Join avn [0] (~avn@
15:24:36 Part pyro_maniac ("Leaving.")
15:25:09pamaurygevaerts: just for information, your last commit to usb-storage put a break before a portion of code which is between #if 0 #endif. If someone changes it to #if 1 then the code is between a break and a case and is never executed
15:25:53gevaertshm, right
15:26:00*gevaerts might fix that later
15:26:56pamaurygevaerts: I found out the "bug" about controls which have associated usb string
15:31:30 Quit DataGhost (Ping timeout: 264 seconds)
15:32:25 Join DataGhost [0] (
15:32:25 Quit DataGhost (Changing host)
15:32:25 Join DataGhost [0] (~dataghost@unaffiliated/dataghost)
15:33:37*FlynDice has taken cover off vegetative clip+ and looks for points to short for miraculous recovery procedure. Opens mind to inspiring ideas and searches diligently
15:34:42funmanFlynDice: did dumping the block device give you anything useful?
15:35:08FlynDicehow do I dump it?
15:35:29 Join xiainx [0] (
15:35:58FlynDiceso switch the if=/dev/sdd?
15:36:07 Join Luca_S [0] (
15:37:25 Join halmi [0] (~Miranda@
15:38:09FlynDicesorry, do I do dd if=/dev/sdd of=dump.bin for a command then?
15:41:06FlynDiceI'll try that later, out of time now...
15:44:32 Quit MagusG (Ping timeout: 252 seconds)
15:45:19 Quit Luca_S (Quit: CGI:IRC (EOF))
15:47:59 Quit halmi (Quit: halmi)
15:49:40 Quit funman (Quit: free(random());)
15:51:18 Join kugel [0] (~kugel@rockbox/developer/kugel)
15:54:50 Quit froggymana (Quit: CGI:IRC)
15:57:54 Join mikroflops [0] (
15:59:12 Quit ender` (Ping timeout: 276 seconds)
15:59:42 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
16:01:39 Quit mikroflops_ (Ping timeout: 240 seconds)
16:09:03 Quit pamaury (Quit: Quitte)
16:13:01 Join ender` [0] (
16:19:38 Join DataGhost_ [0] (
16:19:39 Quit DataGhost (Disconnected by services)
16:19:39 Nick DataGhost_ is now known as DataGhost (
16:28:10 Join halmi [0] (~Miranda@
16:29:25 Quit JohannesSM64 (Quit: WeeChat 0.3.2-dev)
16:29:58 Quit halmi (Quit: halmi)
16:31:04 Join JohannesSM64 [0] (
16:33:40 Join halmi [0] (~Miranda@
16:35:05 Join ssorgatem [0] (
16:36:27 Join yorick [0] (
16:36:47yorickI think it would be nice to have the tracker in fact use flyspray priority feature
16:37:01yorickthen I can see if there are any serious(limiting) bugs in rockbox trunk
16:37:30Bagderbut what good what that bring?
16:37:48yorickprioritizing things is great
16:38:14gevaertsYes, and we all prioritise bugs, just not in a shared list :)
16:38:17Bagderyes, I guess it can be good
16:38:29gevaerts*all* bugs are limiting
16:38:33yorickwouldn't it be better to have a shared list?
16:38:34Bagderif there was a way for us to do that prio
16:39:06 Join funman [0] (~fun@rockbox/developer/funman)
16:39:23funmanssorgatem: hi
16:39:24gevaertsyorick: the importance of a bug depends on the person looking at it
16:39:37ssorgatemfunman: hi
16:40:02yorickgevaerts: but "ipod video not charging with your apple(c) charger(tm) which cost 30 euros" is a bit more limiting than "ipod video scrolling not working until key clicked"
16:40:21funmanssorgatem: i look at a patch I can make you try without bricking the fuzev2
16:40:23gevaertsyorick: only for people who have such a charger
16:40:55ssorgatemfunman: maybe disabling OF autoboot when a usb detection is detected?
16:41:02yorickgevaerts: any usb charger does it :)
16:41:03Torneyorick: a large number of people seem to not have even noticed the charging issues, but several of them have still been rather annoyed by the scrollwheel bug :)
16:41:13funmanssorgatem: that will work only if the second detection (button) is working
16:41:18Torneyorick: so i suspect a lot of people would disagree.
16:41:22yorickthe scrollwheel bug is in a stable release
16:41:25yorickthe charger thing is not
16:41:36linuxstbyorick: And it's only important for people who use an ipod video...
16:41:45Tornesure it is; if you press the button to do charging only, it doesn't charge much/at all ;)
16:41:47gevaertsyorick: would increasing the priority of that bug make people suddenly work on it?
16:41:51ssorgatemfunman: mm and we have no way to see if it's working now, right?
16:41:54TorneThis is kinda deceptive with the name, no? :)
16:42:02Torne(charging mode doesn't charge, the othe rmode does? :)
16:42:28yorickTorne: who presses that button :P
16:42:29kugelfunman: has someone tested if the i2c detection also works on as3525v2?
16:43:03Torneyorick: lots of people? :)
16:43:16funmankugel: it doesn't work in normal build, testing in mkamsboot is looking for problems
16:43:23gevaertsyorick: bug priorities work if you're in an environment where you can tell people what to do and fire them if they don't. I don't think they work in an environment where people do whatever they like
16:43:25Torneyorick: but seriously, anyway. it was just an arbitrary example
16:43:32kugelssorgatem: you should try a delay instead of disabling completely
16:44:00Torneyorick: priorities aren't very useful to the developers for the reason gevaerts says; priorities aren't very useful to users because different users will have vastly different ideas about what bugs are important
16:44:13yorickok then
16:44:18TorneTher eis one thing we could do with the tracker that might help, though
16:44:26TorneThe affected player option should be a multiselect really
16:44:28funmankugel: how does look ?
16:44:34ssorgatemkugel: it's to some of you givving me a patch which I'll hopefully compile and install
16:44:43Tornebecause there's currently no way to say "all PP5020 targets" or "all ipods" or "all CF targets"
16:44:47Torneor similar
16:44:54Torneso it's hard for people to see which bugs *might* affect their player
16:45:00linuxstbTorne: With the problem that the original poster can only say for sure that it's on his/her target...
16:45:17yoricklinuxstb: you can change stuff
16:45:18ssorgatemkugel: my coding skills stop at perl/python
16:45:19Tornelinuxstb: sure, but I mean from the developer's POV
16:45:25Torneyorick: we can, but users can't :)
16:45:33yorickTorne: lets change that :P
16:45:43Tornelinuxstb: a lot of bugs are filed by people who *can* say for sure what targets are affected, also
16:45:47Tornelinuxstb: because they know the code :)
16:45:47linuxstbTorne: I just mean that we don't want users (or even devs) guessing...
16:46:01kugelfunman: looks ok to me, except you're wasting a cycle there :p
16:46:05linuxstbTorne: But they don't know for *sure*...
16:46:07TorneI guess the thing to do would be have a single-select for "reported on"
16:46:12Torneand multi for "confirmed to affect"?
16:46:17funmankugel: where?
16:46:18Torneor is that too complkicated to be useful.
16:46:19Tornei duno.
16:46:29 Join toffe82 [0] (~chatzilla@
16:46:33 Join M3DLG [0] (~M3DLG@
16:46:40kugelyou could bne boot_of and fall through to wait
16:46:50kugelbut it really doesn't matter
16:47:04funmanssorgatem: try , it should always boot the OF, but with some delay inserted if you press the left key or not
16:47:05Tornelinuxstb: *shrug8 it was just a thought
16:47:55funmanthe delay might be a bit long (10s of seconds) though, it's better than too fast to notice
16:48:22ssorgatemfunman: ok
16:49:28linuxstbTorne: Are you sure flyspray allows us to make it a multi-select list?
16:49:52Tornelinuxstb: i have no idea
16:49:59Tornethis was entirely hypothetical :)
16:50:38Torneyorick: i assume what you're after is "i have an ipod mini 2g, is the current build likely to work well enough for normal use"
16:50:50Tornethe intention is that for the ports marked stable, this is *always true* :)
16:50:50yorickexcept I have an ipod video 5.5g
16:50:55Torneyes, it was arbitrary
16:51:06TorneMy point is that the stable ports are, generally, stable.
16:51:07yorickbut I don't want stable
16:51:10yorickI want relatively stable
16:51:15TorneNo, I don't mean the releases
16:51:20TorneI mean, the categories on the front page
16:51:40TorneThe stable ports are, generally, stable in svn *all the time*, with occasional exceptions because someone made a mistake
16:52:06funmanif you don't want stable we could probably add some bugs, just open a feature request
16:52:25Torneipod video 5.5g is also my primary device
16:52:39Torneand yes, I expect that *any* random svn version i happen to pull on any particular day works perfectly fine
16:52:48TorneIn a couple of years this has, so far, always been true
16:52:52Torneat least for any feature I see or use
16:52:59ssorgatemfunman: upgrading firmware
16:53:30TorneFor the unusable targets, this is different, of course :0
16:53:41funmanssorgatem: try first with left pressed, it should boot instantly
16:53:43Torneplayback might work okay in only a small range of versoins, or whatever. but they're listed as unusable for that reason :)
16:54:08ssorgatemfunman: it does
16:54:14funmancool at least it's not bricked
16:54:27Torneyorick: from that point of view, *all* the bugs that affect ipod video are probably fairly low priority. The charging issue has a trivial workaround and for most people, it does still charge, just very slowly.
16:54:39ssorgatemfunman: and does the same without pressing anything
16:54:42Torneyorick: it doesn't make the player unusable, or even unreliable
16:55:02*kugel was about to say the usb detection is more likely to fail
16:55:06funmanssorgatem: ok, try increasing the value on the wait: line, add a 0 (to make it 16 times bigger)
16:55:36funmanssorgatem: hm wait
16:55:40funmandid you run make in dualboot/ ?
16:56:15ssorgatemfunman: no, make clean && make in mkamsboot/
16:56:18funmanyou need to do: cd rbutil/mkamsboot; make -C dualboot ; make
16:56:44Torneyorick: is this not your experience? :)
16:57:16funmanwe have a pre-built dualboot object code in svn, so arm cross compiler isn't needed to build mkamsboot. And this code isn't automatically rebuilt if you only build mkamsboot itself
16:57:16yorickTorne: not yet :P
16:57:20Torneyorick: also, *are* there any current bugs that you would consider high priority for ipodvideo, because they actually significantly impede your ability to use it?
16:57:22yorickI just want to be sure :)
16:57:46yorickTorne: the charging bug?
16:57:46linuxstbThere was an intention at some point that the "TargetStatus" page should list those kind of major issues for a device.
16:57:49Tornewell, that's the idea, anyway. we are human (well, most of us), but we make every effort to not break stuff that already works
16:57:57ssorgatemfunman: rebuilt
16:58:26Torneand if something important is broken in the course of doing something else, and it's not trivial to fix it, we're not going to just open a bug and leave it, we are going to revert whatever change broke it, and try again later, generally.
16:58:38linuxstbyorick: Have you seen this table?
16:59:18ssorgatemfunman: upgrading firmware
17:00:02ssorgatemfunman: the same
17:00:09funman Clipv1 poll report: 5 out of 6 people never experience crashes, 1 experience sparse crases
17:00:19 Join {phoenix} [0] (
17:00:25funmanssorgatem: can you upload your patched OF somewhere so I can check it ?
17:00:32yorickI experience crashes
17:00:37yorick(well once)
17:00:38ssorgatemfunman: ok for example
17:00:45funmanyorick: on Clipv1 ?
17:00:51ssorgatemfunman: where could I upload it?
17:01:14yorickfunman: no
17:01:49 Quit Zagor (Quit: Leaving)
17:02:25Torneyorick: i've done most of the usb power rework required to start fixing charging, btw ;)
17:02:32 Join fyrestorm [0] (
17:02:45Torneyorick: need to finish poking the code to handle when it decides the attached thing is probably a charger
17:02:57Tornethen get various people to test tha twe reaally are flipping the right GPIOs for all the different ipod models
17:03:06 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
17:03:12Tornebut i can commit the different models one at a time, if needed; I know the ipodvideo one works :)
17:03:32Tornethe code i have on my ipod atm detects USB hosts correctly and charges from them, it just doesn't work with AC chargers still
17:04:53Torneso it is being worked on, i just don't have infinite free time ;)
17:04:56ssorgatemfunman: I'm uploading it. It will take some minutes, as it's 5,7 Mb (tar.bz2)
17:06:29***Saving seen data "./dancer.seen"
17:13:28 Quit halmi (Quit: halmi)
17:14:24*yorick goes looking for his Apple iPod(c) Radio Remote(tm)
17:14:57 Join halmi [0] (~netbook@
17:15:02 Quit Xerion (Quit: )
17:15:05 Quit halmi (Client Quit)
17:15:29 Join halmi [0] (~netbook@
17:18:02 Quit M3DLG (Ping timeout: 252 seconds)
17:22:42yorickhmm I can't find it
17:22:51yorickI have probably lost it
17:28:51ssorgatemfunman: uploaded, but not in omgloader
17:31:30 Nick Ypsy is now known as YPSY (
17:33:09funmanssorgatem: strange the USB check is still there
17:33:41ssorgatemfunman: what does that mean?
17:33:53funmanmy patch should have removed the USB detection
17:35:14funmandid the diff apply properly ? can you paste the output of 'svn diff' ?
17:36:15ssorgatemfunman: yes, it applied cleanly
17:36:24ssorgatemdo i paste the output here?
17:36:37funmanno, on
17:36:58funmanah sorry, it's ok in fact
17:37:06funmani misread
17:37:19funmanso the failing check is the button one
17:37:40kugelbut then the delay should come?
17:37:45 Quit avn (Ping timeout: 246 seconds)
17:38:04ssorgatemfunman: here it is:
17:38:26ssorgatemthere's no delay at all
17:38:31funmankugel: true, perhaps the delay is too small to be noticed
17:39:18kugelit's only 10M cycles afterall
17:39:32ssorgatembut if the failing check were the button one, shouldn't it boot into rockbox?
17:39:38ssorgatemor at least try it?
17:39:41 Join avn [0] (~avn@
17:40:10funmanssorgatem: svn revert . -R and then apply this :
17:40:34funmanssorgatem: it could be that dualboot code thinks the left button is always pressed
17:40:50funmanssorgatem: with this other patch you should see a delay when booting normally, no delay when booting by plugging USB cable
17:41:11kugelthe delay should be a bit higher
17:41:28funmani made it 16 times higher than previousy, is that enough ?
17:41:31kugelI assume you need at least 0x20000000 to be noticeable
17:41:59 Quit xiainx (Ping timeout: 246 seconds)
17:42:20kugelmy simple calculation is delay number * 2/250MHz, not sure if it makes sense
17:43:34 Quit MuscleNerd (Ping timeout: 276 seconds)
17:44:16 Join MuscleNerd [0] (
17:44:22ssorgatemupgrading firmware
17:45:45ssorgatemi think there's now a little delay when powering on without usb attached
17:45:51ssorgatembut very little
17:46:02funmanssorgatem: add a 0 to the delay on the wait: line and verify
17:46:53funmangiven that the button code in rockbox driver is a bit more complex than the one in dualboot.S that makes sense
17:46:59 Join xiainx [0] (
17:50:12Bagderrockbox bios! ;-) =>
17:50:14ssorgatemupgrading firmware once again
17:51:59ssorgatemthe same again
17:52:13funmanis the delay more noticeable ?
17:52:44TorneBagder: ick, ugly generated blog linkfarm thing
17:52:49 Quit halmi (Quit: halmi)
17:53:00 Join halmi [0] (~netbook@
17:53:10funmanssorgatem: try with mvn r0, #0
17:53:26*Bagder didn't noticed, but I guess you're right
17:53:27funmaninstead of mov r0, #0x5...
17:53:52ssorgatemso, what instead of what?
17:54:48 Quit Bagder (Quit: It is time to say moo)
17:54:52ssorgatemthe line should be like this: "wait: mvn r0, #0"?
17:55:26 Quit xiainx (Ping timeout: 245 seconds)
17:57:25ssorgatemupgrading firmware
17:59:16funmanstill boots fastly with or without USB plugged?
17:59:49funmanyou ran make in dualboot/ or used make -C dualboot/ right?
17:59:53ssorgatemi'm starting to think i'm the owner of the invincible fuze lol
17:59:56 Quit TheSeven (Read error: Connection reset by peer)
18:00:00 Quit petur (Quit: *plop*)
18:00:18ssorgatemi ran exactly : "make clean && make -C dualboot/ clean && make -C dualboot/ && make"
18:00:27 Join xiainx [0] (
18:01:02funmanhm it could be that both button & USB detection do not work
18:01:11funmanthen you're lucky it defaulted to boot OF
18:02:09ssorgatemah, well, not everything i lost then
18:02:20funmantry to comment out the "bne boot_of" before the "mvn", prefix it with a @
18:02:24ssorgatembut, why is my fuze so special?
18:02:46funmanthis will run the delay unconditionally, then when it's booted you can update to unpatched OF to remove the delay. This will make sure our code is run
18:02:53funmanssorgatem: ask sandisk, not us!
18:03:24ssorgatemfunman: true, it was more like a rhetorical question
18:03:46funmanssorgatem: paste your diff to dualboot.S before running mkamsboot to make sure it's correct
18:04:09funmanssorgatem: and a rhetorical answer :)
18:05:10 Join komputes [0] (~komputes@ubuntu/member/komputes)
18:05:34funmanlooks correct
18:05:50funmanthe delay might be quite long though (still under 1 minute i believe)
18:05:59funmanso don't be afraid
18:08:22ssorgatemupgrading firmware
18:09:11ssorgatemi'm hopefullly seeing a delay
18:09:19ssorgatembecause i don't see anything
18:09:46funmansounds good so far
18:10:33ssorgatemi'm starting to worry
18:11:34funmaneach loop takes 4 cycles, there are 4 billions loops => 16 billions cycles, and the CPU should run at 240 millions cycles per second or less, so it will take 1 minute or more
18:12:11funman1 cycle for sub, 3 cycles for the branch
18:12:19kugela branch is 3 cylces?
18:12:33funmanyes because pipeline must be flushed
18:13:13ssorgatemmore than 2 minutes?
18:13:30funmani don't know the CPU speed at this point though, if it's 24MHz, then it would take a bit more than 10 minutes
18:14:00kugelor maybe it's just bricked :)
18:14:09*funman slaps kugel
18:14:17kugelssorgatem: don't worry
18:14:43ssorgatemso, suposing it's not bricked
18:14:52ssorgatemwhat would be the next to do?
18:15:58funmanfirst upgrade it with OF to remove the (annoyingly long) delay
18:16:26funmanthen reduce the delay to have a good value (some seconds so it can be noticed and not bother too much)
18:16:45funmanand then try to read USB status or a button in a reliable way
18:16:57funmanyour only output is the delay presence or not
18:17:25 Quit JohannesSM64 (Read error: Connection reset by peer)
18:17:45 Join phanboy4 [0] (
18:18:21ssorgatemah, it booted
18:20:05funmanaround 10 minutes then
18:20:32 Quit halmi (Quit: halmi)
18:20:38funman0x2000000 for the delay should be approximately 5 seconds
18:21:05ssorgatemand where do I put that?
18:21:06 Join halmi [0] (~netbook@
18:24:43funman <- this patch disables USB check for the Fuzev2, create a section specific for it
18:25:06funmanand checks C4 pin (select button) instead of C3 (left)
18:25:08 Quit xiainx (Ping timeout: 240 seconds)
18:26:00funmanideally you would read button-fuzev2.c and figure out what's missing in dualboot.S ; if C4 doesn't work I'll try to make a patch later
18:30:37funmani'm leaving, when i make a patch i'll post it on the forum
18:30:38 Quit funman (Quit: free(random());)
18:31:03 Join xiainx [0] (
18:32:00ssorgatemthat last patch doesn't compile
18:42:10 Quit xiainx (Ping timeout: 252 seconds)
18:43:48 Join bmbl [0] (
18:43:53 Quit bmbl (Changing host)
18:43:53 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
18:44:05CIA-5New commit by alle (r25607): Mark const return value; correct the comment about valid menu items for the hotkey
18:46:26TorneJdGordon_: did you ever do anything more with the 32/64mb memory detection code for the ipod video?
18:46:46 Join xiainx [0] (
18:48:13 Quit RadicalR (Ping timeout: 265 seconds)
18:48:36CIA-5New commit by alle (r25608): Better dot placement
18:49:18 Join M3DLG [0] (~M3DLG@
18:50:48 Join RadicalR [0] (
18:52:46TorneJdGordon_: can you post the patch as you had it somewhere?
18:54:47 Quit krazykit (Read error: Connection reset by peer)
18:55:23 Join krazykit [0] (
18:56:21 Quit komputes (Ping timeout: 276 seconds)
18:57:17 Join fml [0] (
18:59:00 Join petur [0] (~peter@rockbox/developer/petur)
18:59:36fmlRe. r25607: it should be possible to extend the types of possible hotkey'able menues but the menu API is done in a way that makes it hard. The reason is that to get the text of a menu item it's necessary to know the index of the item in its parent's menu (which shouldn't be necessary IMO)
19:06:32***Saving seen data "./dancer.seen"
19:11:25 Quit flydutch (Quit: /* empty */)
19:15:01 Quit linuxstb (Ping timeout: 258 seconds)
19:16:45 Join GeeksHaveFeeling [0] (~meow@unaffiliated/ghf)
19:17:18 Quit GHF (Read error: Operation timed out)
19:19:47 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
19:20:08 Nick fxb is now known as fxb__ (
19:20:31 Join webguest99 [0] (
19:21:23 Join G_H_F [0] (
19:21:28 Quit webguest99 (Client Quit)
19:22:08 Join Blue_Dude [0] (
19:22:24 Quit M3DLG (Ping timeout: 260 seconds)
19:22:54 Quit phanboy4 (Quit: Leaving)
19:24:44 Quit GeeksHaveFeeling (Ping timeout: 260 seconds)
19:25:08 Quit TillW (Quit: This now concludes our broatcast day.)
19:25:20 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
19:25:30Blue_DudeI just wanted to throw an idea out there re: the volume clipping/overboost discussion. This is an interface idea: why not have a built-in "pause" in volume control at 0db, when setting volume. That is, make the 0db point a kind of detent you have to step over in order to select something higher than 0db? If we want 0db to be a kind of soft limit, it makes sense to make it easier to stop there.
19:25:47 Join jgarvey [0] (
19:26:32Blue_DudeThis could be done by making the volume range from MIN to 0, 0, 0 to MAX. Two or three chances to hit 0 on the way past to MAX.
19:26:51Blue_DudeAnd easy to reduce to exactly zero on the way down.
19:26:57 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
19:30:00 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
19:33:04 Join tvelocity [0] (
19:34:18kugelBlue_Dude: if we wanted to protect the users so badly we would have done it by now
19:34:22 Quit saratoga (Ping timeout: 248 seconds)
19:35:11Blue_Dudekugel: it's not so much a protection idea as it is an ergonomic feature. It's just an idea I stole from other gear.
19:35:57 Join funman [0] (~fun@rockbox/developer/funman)
19:36:06Blue_DudeBut perhaps more trouble than it's worth. I personally seldom try to hit 0dB exactly but maybe that's important to others.
19:36:50CIA-5New commit by Buschel (r25609): Replaygain pre-amp can be set in 0.5 dB steps.
19:37:08funmanssorgatem: try
19:38:19 Join Buschel [0] (
19:39:37 Quit dfkt (Read error: Connection reset by peer)
19:41:44 Join dfkt [0] (dfkt@unaffiliated/dfkt)
19:42:24 Quit Buschel (Ping timeout: 240 seconds)
19:43:22 Join panni_ [0] (
19:43:43 Join Buschel [0] (
19:46:28*FlynDice gives up on clip+ and heads to store, wonders if ranma would like to try his JTAG voodoo magic on said unit?
19:48:35*funman is in remote discussion with a profesional solderer
19:55:16 Join piotrekm [0] (~piotrek@unaffiliated/piotrekm)
19:55:25 Join esperegu [0] (~quassel@
19:56:01espereguI broke the screen of my iriver H120. anyone knows if that is replaceble and where I could buy that?
19:57:19fmlBlue_Dude: hello. Have you analyzed remote keymaps during your hotkey work?
19:57:56Blue_Dudefml: I didn't even look at them since there were no keys I could safely steal.
19:58:48fmlBlue_Dude: aren't there the "quick" keys to view the playlist? I haven't looked though.
19:59:22Blue_Dudefml: I don't think so, but I didn't look all that closely. There's no reason it wouldn't work though.
19:59:44pixelmathe playlist viewer shortcuts were a mere guess too
20:00:36fmlpixelma: but we could make whatever was "playlist view" to be "hotkey"
20:01:29funmanesperegu: I don't think it's related to rockbox. Better come on #rockbox-community (or try mysticriver website?)
20:01:56 Quit Boldfilter (Ping timeout: 246 seconds)
20:02:11esperegufunman: k. thx
20:02:17 Join Boldfilter [0] (
20:03:06Buschelsaratoga: Even in the asm version replaygain is 1 smull, 1 mov and 1 orr per sample = 4+1+1=6 cycles per sample. Plus and 1 ldm + 1 stm per 2 samples = 3+1+4+2 = 10 cycles per 2 samples. Plus 1 sub and 1 branch per 2 cycles = 1+3 = 4 cycles per 2 samples => 13 cycles per sample = 1,1 MHz (@44.1kHz sample rate). This equals ~0.3 mA and is measurable in terms of runtime.
20:03:27pixelmafml: yes sure but the playlist view wasn't implemented very accurately either, I wouldn't be surprised if no-one took care of remote keymaps and I already know that the chosen combo won't work on the Iaudio M3 btw. (it just existed for a while and no-one noticed, lack of documentation is one cause and the M3 is probably an "exotic" target)
20:04:18 Join wodz [0] (
20:04:35Buschelsaratoga: but of course such savings are minor comapred to other settings.
20:05:37pixelmathere probably won't be a way similar to the other Iaudios, they all have in common that the only combos electrically possible are ones with the "Power" key which also serves as a hard power-off after a rather short time
20:05:42 Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
20:06:12pixelmafml: the playlist viewer shortcut was dropped on the M5X5 pad after I said that here
20:07:18 Quit CGL (Quit: Saliendo)
20:07:40fmlpixelma: OK then. But shouldn't we then remove the remote key for the playlist viewer where it's still present? It's not properly assigned and not documented.
20:08:26pixelmaI'd rather say implement if possible and document
20:09:16pixelmaor I misunderstood
20:09:55pixelmafml: which remote key did you mean, I thought you said that there wasn't one for remotes?
20:12:25 Quit Boldfilter (Ping timeout: 260 seconds)
20:12:45ssorgatemfunman: Now it works
20:12:55ssorgatemfunman: wel, it waits, i mean
20:14:36funmanregardless if you press center button or not ?
20:14:37fmlpixelma: I mean that if a remote key is defined then it should be (temporarily) removed. And then chosen and documented properly.
20:15:12fmlpixelma: I see that H100 has a key for ID3
20:15:26 Join komputes [0] (~komputes@ubuntu/member/komputes)
20:15:28ssorgatemfunman: yes
20:15:34ssorgatemfunman: well
20:15:42ssorgatemfunman: haven't tried pressing the button
20:16:25ssorgatemfunman: pressing the button doesn't wait
20:16:38ssorgatemfunman: that's how it's supposed to be, right?
20:16:51pixelmafml: the ID3 viewer is not the playlist viewer shortcut though (and exist for a longer time)
20:16:52funmanhm that's cool
20:17:19funmanssorgatem: if i give you a workaround to boot rockbox with center button as dualboot, will you still help to find how to read left button?
20:17:40ssorgatemfunman: of course
20:18:17fmlpixelma: then we can make it hotkey. The user can choose it to be ID3.
20:19:17pixelmain that case it should be the former ID3 screen thing everywhere IMO,
20:20:09funmanssorgatem: should default boot to rockbox, boot to OF if center is pressed
20:20:19pixelmanot like "oh, on the remote we don't need the ID3 screen button anymore" while it would be somthing else on the main unit
20:21:15pixelmajust my first thought
20:23:08ssorgatemfunman: upgrading firmware
20:23:10fmlpixelma: maybe
20:23:28pixelmadoes the code even handle two different hotkeys?
20:24:28ssorgatemfunman: it works, but the oder way around
20:24:37notlisteningfunman FlynDice's clip+ still knackered?
20:24:47 Join Boldfilter [0] (
20:24:49fmlpixelma: you mean WPS and tree?
20:24:50ssorgatemfunman: it defaults to OF, but when I press center, it tries to load rockbox
20:25:02pixelmafml: no, remote and main
20:25:34ssorgatemfunman: and fails with "ATA error: -2"
20:25:43funmanssorgatem: hm sad
20:26:01fmlpixelma: it doesn't handle buttons AFAIU, it handles actions which are defined in keymaps
20:26:03pixelmafml: which could potentially make it four different hotkeys then IIUC
20:26:10funmanssorgatem: bootloader might be broken in svn though
20:26:28fmlpixelma: yes
20:26:37pixelmafml: but it stores some settings for them, no?
20:26:51funmanssorgatem: I have r25377, can you build a bootloader from this revision ?
20:28:32ssorgatemfunman: mm how would I do it? I've never downgraded a svn tree
20:29:53fmlpixelma: no, I think it only stores two settings (WPS and browser), but not separately for main and remote
20:30:01funmanhm svn help revert doesn't help. I use git, not svn so I don't know but I hope someone else here does
20:30:51linuxstbssorgatem: I think just "svn update -r XXXXX"
20:31:06linuxstbrevert is for removing local changes
20:31:51ssorgatemlinuxstb: thanks, it seems to have worked
20:32:12ssorgatemit says "At revision 25377", at least
20:33:24ssorgatemwhich is the preferred chainload, arm-elf or arm-elf-eabi?
20:33:41pixelmafml: so currently - if you would have a hotkey action on the remote which replaces the ID3 one and a hotkey action for the main keypad which replaces the playlist viewer shortcut (and assuming that the ID3 action on the main unit still exist) - if you then set your hotkey function to ID3 viewer you would have two buttons (or combos) doing the same on the main kaypad, or do I overlook something?
20:34:10pixelmaor keypad
20:34:15funmanarm-elf-eabi should work, but we still use arm-elf as default (and it's what my bootloader uses)
20:34:53 Join roflmao [0] (~yep@
20:35:10fmlpixelma: I think you're right. And I also think that we should remove a special ID3 button if it exists
20:35:34 Part roflmao
20:35:59pixelmait's there on (almost?) all target and been there for ages
20:36:46 Join kugel_ [0] (
20:37:00 Quit kugel (Disconnected by services)
20:37:04 Nick kugel_ is now known as kugel (
20:37:08 Quit kugel (Changing host)
20:37:08 Join kugel [0] (~kugel@rockbox/developer/kugel)
20:38:58*kugel just saw his assessment task
20:39:55kugelI actually planned to do that anyway, but way later :p
20:40:03fmlpixelma: but then we have an ideal key for the hotkey! Or is it extensively used?
20:40:38 Quit kugel (Remote host closed the connection)
20:40:42fmlpixelma: I mean the old ID3 key
20:40:49 Join kugel [0] (~kugel@rockbox/developer/kugel)
20:40:59ssorgatemfunman: let's see if this time it works
20:41:00pixelmanot by me but then 2 of my 3 targets don't have it. People using both might be opposed
20:41:27pixelmaID3 screen and playlist viewer shortcut I mean
20:41:57fmlpixelma: I don't think there are many people who use both since hotkey was introduced just recently
20:42:06ssorgatemfunman: interesting
20:42:22pixelmafml: not hotkey but the playlist viewer shortcut
20:42:32ssorgatemfunman: rockbox now gets further
20:42:46ssorgatemfunman: there's a quic error i can't read and then
20:42:49pixelmafml: which wasn't there for too long but already a while
20:43:05funmanssorgatem: do you have .rockbox folder on the player ?
20:43:08ssorgatemfunman: a screen with GPIO values and directions
20:43:32funmanssorgatem: hm that means an ATA error as well I think. Do you have a µSD card plugged in ?
20:43:52ssorgatemfunman: no µSD card
20:43:55 Join Strife89 [0] (~michael@
20:44:09ssorgatemfunman: I think i have the .rockbox folder
20:44:27 Join einhirn [0] (
20:44:29ssorgatemfunman: *PANIC* ata -2
20:44:42ssorgatemfunman: and after that loads the OF
20:44:43kugelwhat compiler do you use?
20:44:59kugelthe eabi version by chance=
20:45:00funmanwhich means timeout in initialization step
20:45:06fmlpixelma: we can still change that. Keymaps are not cemented. It's *soft*ware :-)
20:45:20ssorgatemrockbox may be compiled with the eabi
20:45:37kugelfunman: the delay func in sd-as3525v2.c fails with the eabi toolchain because it's optimized away
20:45:49pixelmafml: ask Llorean about changing keymaps ;) Although my experience wasn't that bad
20:46:04kugelsince it's not volatile
20:47:34fmlLlorean: how are you about changing keymaps? (See discussion about "Show ID3" button just above)
20:47:37ssorgatemi'm trying now with latest build from
20:47:52kugelssorgatem: I assume the bootloader is failing as well
20:48:07Lloreanfml: I think pixelma was just mentioning how much flak I got when I changed a few keymaps way back.
20:48:12funmanssorgatem: rockbox (from .rockbox/ directory) isn't loaded afaiu
20:48:12ssorgatemit loads the menu for a fraction of a second, then loads the OF
20:48:34funmancan you see the screen flicker? you would see the version number (svn revision + date) change
20:48:46funmanah so the bootloader works
20:48:50pixelmafml: you reduce the number of used buttons again though and then people complaining about an unused button and in the worst case want a second hotkey
20:48:58ssorgatemit works
20:48:59LloreanI wouldn't know how may targets have a "Show ID3" button, I've never used it on any of mine or even accidentally hit it I think
20:49:05ssorgatemi'm ion rockbox
20:49:35fmlBTW: how do you people feel about not displaying the splash when the hotkey is not assigned?
20:50:10funmanssorgatem: nice! do all buttons work ? (especially left button)
20:50:24funmanand what happens if you plug the USB cable while in rockbox ?
20:50:41 Join Luca_S [0] (
20:50:56linuxstbfml: I was wondering about that feature - it serves as a way to tell users that they can assign a hotkey to it, so is useful for that purpose... But I imagine it could be annoying if you normally use that button to enable the backlight...
20:51:06*linuxstb adds the disclaimer that he hasn't updated Rockbox recently...
20:51:40ssorgatemall buttons work
20:52:01ssorgatemi'm trapped in solitaire
20:52:23funmantrapped? :)
20:52:31ssorgatemno, it's ok
20:52:35funmanssorgatem: try the moving the red queen
20:52:36funmanah ok
20:52:37Luca_Suse a long press of the home button
20:52:41fmllinuxstb: yes, that's what I'm thinking. Many have resisted the rec key assignment to any function because they already see "backlight on" as a function
20:52:42ssorgatemi've to press the home button
20:52:44ssorgatemmore than once
20:53:00fmllinuxstb: ...which deserves its own key
20:53:47ssorgatemthe fire demo looks weird
20:53:49linuxstbfml: Unless it can be either "unassigned" (in which case, you get the splash), or "do nothing" (in which case you don't)...
20:53:53funmanssorgatem: keep the bootloader you built around, i'll try to see what happens in button code (unless kugel beats me at that)
20:54:05ssorgatemfunman: ok
20:54:14funmanssorgatem: hm and what happens if you plug USB ?
20:54:53fmllinuxstb: that I didn't understand. Could you elaborate?
20:54:54funmanif the previous tests with mkamsboot/dualboot were correct, USB should always be detected (and rockbox would reboot immediately)
20:55:05funman, even if there's no USB connection
20:55:06ssorgatemexiting froma plugin and then having a black lcd was a known issue, roght?
20:55:17Luca_Sssorgatem: yes
20:55:40fmllinuxstb: ah, you mean the function "do nothing"!
20:55:49ssorgatemlet's see what happens with usb
20:56:13fmllinuxstb: hehe, we could add that function!
20:56:20ssorgatemif i can get into rockbox again
20:56:31ssorgatemit flickers, then loads OF
20:56:54 Quit ender` (Ping timeout: 264 seconds)
20:57:05funmanhum, i'm lost there
20:57:42ssorgatemif I hold the select button
20:57:42funmanOF should only load if you press the center button
20:57:50ssorgatemuntil the rockbox menu
20:57:56ssorgatemit remains in rockbox
20:58:12ssorgatemOF loads if I do not press the center button
20:58:28funmanah sorry
20:58:33kugelgevaerts: quite a demanding job for a assessment task isn't it?
20:58:49funmanso left button / USB detection works fine in rockbox, but not in dualboot, unless we messed the check for USB
20:58:56ssorgatempluging it in the usb adapter does nothing
20:59:05gevaertskugel: maybe :)
20:59:22funmanah ok, USB detection doesn't work either in rockbox
21:01:18 Quit krazykit (Ping timeout: 252 seconds)
21:03:55ssorgatemmm, trying to launch doom gives an "undefined instruction" white screen, then loads OF
21:04:04 Join krazykit [0] (
21:06:11funmanssorgatem: did you see the status page ?
21:06:23ssorgatemfunman: yes
21:06:36***Saving seen data "./dancer.seen"
21:06:49ssorgatemfunman: but should it load OF?
21:07:30funmanwhen rockbox 'panics' it show a message on screen (could be the white screen instead if something bad happened), and reboot on keypress
21:07:53funman(not all models reboot on keypress though, but the fuze does)
21:08:52 Quit esperegu (Read error: Connection reset by peer)
21:09:37 Join MethoS- [0] (~clemens@
21:12:08 Join ender` [0] (
21:19:21 Join esperegu [0] (~quassel@
21:19:57espereguI get a 404 when trying to install voice. it seems that the date is not correctly formated:
21:21:43espereguis there a way to manually fix that?
21:22:49fmlJdGordon_: ping
21:24:08CIA-5New commit by tomers (r25610): Text viewer: Remove unused max_line_len variable
21:24:35 Join pamaury [0] (~c2c7a50a@rockbox/developer/pamaury)
21:24:48 Join flydutch [0] (
21:24:56linuxstbesperegu: What do you mean? Are you using Rockbox Utility?
21:25:29CIA-5New commit by tomers (r25611): Text viewer: Make calculation clearer
21:26:28esperegulinuxstb: yeah
21:26:39esperegulinuxstb: I think it has the wrong url
21:26:59gevaertsesperegu: which version of Rockbox Utility is that?
21:27:09linuxstbesperegu: I don't know about that, but to install that zip file, you simply unzip it to your device. It already contains the correct directory/folder structure.
21:27:19 Join tomers_ [0] (
21:27:19esperegugevaerts: just downloaded it
21:27:28esperegulinuxstb: in the root.
21:27:30gevaertsah, right...
21:27:40esperegugevaerts: it should get it from
21:27:47linuxstbesperegu: Yes. The zip should have ".rockbox" as the top-level directory, which is located in the root of your h120.
21:27:54 Quit toffe82 (Read error: Connection reset by peer)
21:27:57esperegubut it tries
21:28:08gevaertsesperegu: 1.2.6 is ready to be released, but apparently it's not on the servers properly yet. That one should fix this issue
21:28:43CIA-5New commit by tomers (r25612): Text viewer: Fix FS #11190 - Text Viewer shows no Text when skipping to last page
21:29:19 Quit einhirn (Read error: Connection reset by peer)
21:29:47espereguI really need voice since my screen is broken :-(
21:29:49gevaertsdomonoky: can I update the RockboxUtility wiki page? It looks like 1.2.6 is on the download server
21:30:04notlisteningfunman did you see my post in the forums about the battery levels?
21:30:42notlisteningon the clip+..?
21:31:15funmannotlistening: yes, did you see my answers ?
21:31:35notlisteninglol thats a no just take a peek
21:31:41 Quit fml (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
21:34:01notlisteningfunman, the behaviour i saw was the OF shutting off when just trying to switch it on which seems weird?
21:34:14notlisteningbut i know you guys know your stuff
21:35:32gevaertsesperegu: the new download links are there now :)
21:35:43 Quit Schmogel (Ping timeout: 240 seconds)
21:36:27 Quit linuxstb (Ping timeout: 258 seconds)
21:38:16funmannotlistening: the OF said it was discharged, and rockbox said the battery was full ?
21:38:38funmanand no I don't know well this part, bertrik knows better than me
21:38:59 Join TheSeven [0] (~theseven@rockbox/developer/TheSeven)
21:41:08notlisteningno they both reported low battery just rockbox seemed to just carry on and on and on and I got a bit worried about damaging the rechard curve of the battery by totally exhausting it
21:41:33 Join Watermark [0] (
21:42:39funmandid you check what % rockbox reported ? in the debug menu > battery screen
21:42:40espereguhmmm.. there is already a voice file there.
21:44:51esperegushould I hear something after putting those files there?
21:45:29notlisteningit was saying 0% but under rockbox info
21:45:41notlisteningI will check next time it is running on thin air ;)
21:47:11*FlynDice admires new clip+4GB and waits patiently for it to charge up so he can begin the torture.....
21:47:31espereguHa! found it. was in the config file set to off
21:47:37esperegumanually changed the cfg file
21:47:42notlisteningFlynDice good news bad news u had to get a new one :) :(
21:48:46notlisteningFlynDice, did you try a full DD of the player one to the other?
21:50:00FlynDiceI can only access a 4MB area so I don't think that's going to work... I'm open to trying something if you've got an idea though
21:50:59wodzamiconn: ping
21:51:07 Quit bertrik (Read error: Connection reset by peer)
21:53:31notlisteningdo you think a full DD image of a working unit would include that 4MB space and does having the correct content in those 4MB give you a working clip?
21:53:45notlisteninghaha what do you have to loose now?
21:54:05kugelhrm, it doesn't seem as straight forward as I thought
21:54:05*gevaerts suspects not
21:54:25espereguwhat is the best tts to use?
21:55:38archivatoresperegu: what OS?
21:55:41 Quit Strife89 (Quit: Clocking out.)
21:55:45espereguarchivator: linux
21:55:51espereguarchivator: kubuntu
21:56:00 Part clauwn
21:56:20archivatoresperegu: I usually go with festival and the cstr_us_jmk_arctic_multisyn voice
21:56:27yorickhmm would rockbox run on an X4-tech clipman?
21:56:46archivatorIt can be pretty slow, though. The nitech_us_* voices are all quite good, too.
21:57:23espereguarchivator: well. I could just let it run overnight
21:57:49archivatoresperegu: not *that* slow (unless you have a 486) :)
21:58:22espereguarchivator: =)
21:58:44espereguarchivator: is that a config option? (cstr_us_jmk_arctic_multisyn)
22:00:16FlynDicenotlistening: I'm trying your dd idea now I'm dumping the new clip+ to a file and I will dd that back onto the broken one. It can't make things any worse....
22:00:53 Join merbanan [0] (
22:01:15 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
22:02:06espereguarchivator: how should I install that voice?
22:02:28yorickI love the thing...touching the aluminium case while it was connected to usb used to disrupt the connection and give you a shock <3
22:02:45yorickdisplay is OLED, and burns in after 15 minutes and broke down eventually :)
22:03:45yorickbut it did hang around your neck conveniently with headphones built in to the cord
22:07:12archivatoresperegu: well, it's not hard. You can get a bunch of extra voices here:
22:07:33archivatorIt's a matter of unpacking the folder in /usr/share/festival/lib/voices-multisyn (in Fedora)
22:08:03archivatorHowever, I believe the Debian packages use a non-standard directory structure.
22:08:11 Join linuxstb_ [0] (~linuxstb@rockbox/developer/linuxstb)
22:09:02 Quit linuxstb (Ping timeout: 258 seconds)
22:09:29 Nick linuxstb_ is now known as linuxstb (~linuxstb@rockbox/developer/linuxstb)
22:10:27espereguarchivator: what's the difference between cmu and cstr ?
22:11:03*gevaerts has this suspicion that festival voices aren't really on-topic here
22:11:55archivatorgevaerts: but voice file building is!
22:12:49 Join fml [0] (
22:13:25fmlBuschel: Hello. Why is the '{}' needed in '0.5{}dB'?
22:15:40 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
22:15:44fmlbluebrother: ah, that ^^ was originally written by you. So why?
22:17:00Buschelfml: I didn't want to change this as I wasn't sure about this, too. Seems to be obsolete.
22:17:11FlynDicenotlistening: good idea but..... badabing no dice......
22:21:10 Quit yorick (Quit: Lost terminal)
22:21:59 Join Schmogel [0] (
22:27:00*TheSeven always likes to fully understand the boot process of a device before tinkering with it
22:27:20TheSeven(including recovery mechanisms)
22:27:52fmlWhat are all those files ending with a ~ in PD box?
22:28:03 Quit Buschel (Ping timeout: 240 seconds)
22:28:12fmlgevaerts: ^^?
22:28:17 Join TillW [0] (
22:28:38 Quit fml (Quit: CGI:IRC)
22:28:39gevaertsfml: those handle waveforms. Blame the PD people
22:28:45 Join fml [0] (
22:29:04 Quit merbanan (Ping timeout: 265 seconds)
22:29:35 Join merbanan [0] (
22:29:41 Join Xerion [0] (~xerion@
22:31:28CIA-5New commit by alle (r25613): Remove confusing braces
22:32:16 Join luistmw [0] (
22:32:54 Quit fml (Client Quit)
22:33:42luistmwcan someone help me
22:33:58krazykitluistmw, not if you don't ask a question
22:34:20luistmwokay so i have an ipod nano 2g
22:34:59luistmwand i tried to uninstall rockbox but i only errased the .rockbox folder
22:35:34krazykitand so you still have the bootloader
22:36:24luistmwwhat can i do to put it back to normal
22:36:52krazykitdoes ipodpatcher not have an "uninstall" option?
22:37:15luistmwyes but my pc wont detect the ipod
22:38:09archivatorluistmw: you'll need to enter disk mode
22:38:42luistmwback and select
22:38:55luistmwand after that?
22:39:32archivatorconnect to the PC, use ipodpatcher to delete the bootloader.
22:39:39linuxstbluistmw: back and select isn't disk mode...
22:40:03archivatorThere seems to be a −−delete-bootloader option but I'll let someone else confirm that it does what you need.
22:40:28linuxstbarchivator: Yes, it does. It's also called "uninstall" if you run ipodpatcher without parameters..
22:40:53luistmwill try
22:41:11luistmwbe right back
22:42:53 Quit esperegu (Remote host closed the connection)
22:44:11Watermarklinuxstb you there?
22:44:50 Quit merbanan (Ping timeout: 248 seconds)
22:47:05 Quit {phoenix} (Remote host closed the connection)
22:47:43 Quit ssorgatem (Remote host closed the connection)
22:49:15 Quit bmbl (Quit: Bye!)
22:50:17luistmwhow do i put it in diskmode
22:50:50TheSevenreset (menu+select) and then immediately select+play when it reboots
22:51:12 Join b0hoon [0] (
22:51:36TheSevenhm, we should add some "press any key to enter disk mode" thing in the bootloader's error handler, instead of just sitting there
22:51:41luistmwthank you
22:52:11TheSevenluistmw: another option would be just booting the apple firmware and uninstalling the bootloader from there
22:52:21TheSevento do that, turn on the hold switch after resetting your ipod
22:52:40b0hoonHi. Will it be a good proceder to commit a part of the manual without the plugin keymaps so far? For me it could ease syncing changes in SVN.
22:52:52TheSeven(during the apple-logo screen, before the bootloader comes up)
22:53:53b0hoonOr it will be better to commit the whole thing at the end?
22:54:12Watermark@linxuxstb, have you found a fix or other options to try to fix my COWON S9 with Tcctool?
22:54:23gevaertsb0hoon: I'd commit early and often
22:54:40linuxstbWatermark: No, I haven't looked.
22:54:51Watermarko ok just checking
22:55:25Watermarki had my S9 plugged in all night via UAB to charge the Batt.
22:55:28b0hoonallright, so no objections?
22:55:52 Quit archivator (Quit: Leaving)
22:56:05linuxstbDid you try what I suggested last night - holding the "on" button whilst uploading the fimware?
22:56:34Watermarkyes, it still just disconnects when patching is complele
22:56:53kugeljhMikeS: ping
22:57:31b0hoonSo /me backs to writing the manual...
23:01:18dfktWatermark, i wonder if the s9 can actually charge when its bricked, and/or when it's in recovery mode - or if that's a catch-22
23:01:43Watermarkidk really
23:02:32 Quit pamaury (Quit: Page closed)
23:02:55Watermarki'm just why the device disconnects. could it be because of the Battery?
23:02:58dfkttelechips of course doesn't have the data sheet available online, just via some passworded login
23:04:14linuxstbWell, there's no proof (afaik) that tcctool will even work with the S9 at all - telechips may have changed things...
23:04:55 Quit S_a_i_n_t (Ping timeout: 276 seconds)
23:05:16luistmwok im back
23:05:35luistmwthanks to all of you who helpedme
23:06:38*domonoky reads the log and notes, that rbutil should also be able to uninstall a bootloader from any ipod.
23:06:40***Saving seen data "./dancer.seen"
23:06:43Watermarktrue linuxstb. but what's the point of Recovery Mode then?
23:07:23 Part luistmw
23:07:36linuxstbWell, it has the same purpose, it may just work differently.
23:08:32Watermarkprobably. Sighs* me stuck with a bricked S9 but not all that bad. could just get the COWON J3 when it comes out
23:08:33linuxstbBut I would guess it's the same, and someone just needs to determing the value for SDCFG (the SDRAM configuration register) that's uploaded to the device as part of the header that tcctool adds to the firmware image.
23:08:40 Join anewuser [0] (anewuser@unaffiliated/anewuser)
23:09:33Watermarkhey dfkt you think dalmane98 could help out?
23:10:40Watermark@linuxstb, how can we fin that Info. out?
23:11:04linuxstbEither guess, find it by disassembling the original firmware, or find a copy of the datasheet.
23:11:37Watermarkhow would we disassemle the FW?
23:12:25 Part b0hoon ("GTG. Bye.")
23:12:50linuxstbOr in some cases, the value has been found because it's been documented somewhere. e.g. I think Cowon made the telechips "fwdn" tool available to help people unbrick some devices (maybe I6/I7), and the instructions for that included that value.
23:13:09linuxstbOr if the manufacture releases an official recovery tool, it can be found via a usb sniffer.
23:15:13CIA-5New commit by tomers (r25614): Text viewer: Fix wrong calculation of bookmark's position (introduced by r25611) ...
23:20:56 Quit bertrik (Ping timeout: 260 seconds)
23:28:24 Quit tomers_ (Ping timeout: 260 seconds)
23:29:44 Quit tvelocity (Quit: Αποχώρησε)
23:31:05 Join yelped [0] (
23:31:37yelpedAnyone have time to check this out?
23:34:32Watermarkcan it be any version of fwdn tool?
23:34:42dfktWatermark, i don't think dalmane can help with that
23:35:18Watermarko nvm
23:35:27dfktlinuxstb, would it be possible finding the right value via trial & error? is it a base-2 value?
23:39:57Watermarkwhat Value are we looking for again. me emailing COWON
23:40:19 Quit notlistening (Remote host closed the connection)
23:40:29funmandfkt: 0x42e97010 0xa2e92010 0x62e97010 0x52e97410 0x62e92018 <- trial & error look a bit hard
23:41:36TheSevenfunman: doesn't look too bad
23:41:52TheSeveni'd say around 50 tries and you're set
23:42:11Watermarkwhat are those numbers? i'm a noob here Lol
23:42:14TheSeventhose values aren't differing terribly much
23:42:41TheSevenfrom what i understood, these are values that have been seen for that SDCFG field
23:42:42funmancommon factors between 2 values: 2, 3
23:42:43*linuxstb tries to find a description of that register...
23:43:38Watermarkwell, i hope we can get this sorted out guys
23:43:46Watermarkneed all the help we can get
23:44:16TheSevenfunman: 4 possibilities for the first digit, 2 for the fifth one, 2 for the eigth one
23:44:55funmanoh that way there is much more common 'factors'
23:45:51TheSevenand if we get a basic understanding of that reg's contents, we can probably just go for some safe-slow-stupid values ;-)
23:46:16 Join einhirn [0] (
23:46:52 Quit anewuser (Quit: Like renoise + like music? 3 days to submit your entry!)
23:47:06 Join Curtman [0] (
23:48:57funmani was looking at the setting of SDRAM mode register in as3525 but it doesn't look similar (lower bits have no signification it seems)
23:49:44Watermarkover at a Russian site i found something on the D2 for unbricking
23:49:49 Join anewuser [0] (anewuser@unaffiliated/anewuser)
23:49:53Watermarkhas some values
23:49:58 Quit komputes (Remote host closed the connection)
23:52:28dfkti have the FWDN version that is compatible with the d2, but that's probably no help for the s9... seeing as the d2 is supported by tcctool already
23:52:56 Join stripwax [0] (
23:52:59dfktfwdn for the d2 says Custom SDCFG=0xA2E92010
23:53:20Watermarkhow'd they get ahold of it?
23:53:31dfktthat would be good to know, indeed
23:54:36Watermarki have a great site. Google is your friend :)
23:54:59Watermarkit's in Russian though
23:55:28 Quit piotrekm (Quit: piotrekm)
23:55:35Watermarkgod detail for the S9
23:56:38dfktnothing that helps, as i see it
23:56:56Watermarkstill though

Previous day | Next day