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

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

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

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

#rockbox log for 2011-01-02

00:00:01saratogagenerally just because two SD cards are packaged the same doesn't mean they're the same internally
00:00:32tuxifiersorrowly :)
00:01:03tuxifierI already even tried to dd the image of the working card to the unrecognized
00:01:08tuxifierno success
00:02:41saratogaprobably just a bug in our controller driver that only occurs with some cards
00:02:58wah_wah_69Hi everyone! I have just made a post about the vibe 300 in the forum, take a look if you're interested in a new port.
00:05:38saratogai really doubt anyone here is interested
00:05:57wah_wah_69Hi everyone!
00:05:57wah_wah_69I've owned a Packard Bell Vibe 300 for aroud 4 years (It was a christmas gift), the first thing I made when it was given to me was to take a look around here for a port, but there was none :'(
00:06:27wah_wah_69crap, this irc client, I pasted all that without wanting
00:06:47wah_wah_69So who might have info to help me develop a new port?
00:08:22saratogahave you seen this: http://www.rockbox.org/twiki/bin/view/Main/NewPort
00:09:15wah_wah_69I'm reading it right now, its quite light on technical details
00:09:35saratogayeah, you pretty much have to figure those our yourself
00:10:30soapthomasjfox, I appear to have forgotten the root password of my N810!
00:10:36wah_wah_69The thing is, having the same portalplayer chip as some supported players it might be more or less easy
00:10:46thomasjfoxsoap: trustno1?
00:10:52saratogai guess a good first step would be to figure out how to recover from a bad flash
00:11:17saratogathen dig through the firmware file and see if you can figure out which GPIO pins go where
00:11:32gevaertswah_wah_69: it's going to be easier than some ports, yes
00:11:39wah_wah_69I don't know this even is flashable, it has an mi4 file in the sytem directory
00:11:47saratogathen try to blink the lights on or something
00:12:10thomasjfoxsoap: On the n900 it usually works by "sudo gainroot" or "root" in the terminal
00:13:09saratogayes thats the "normal" PP method, the H10 and i think Phillips players use that
00:13:37soapoh, thomasjfox, I set a root password years ago. But my "default" root password isn't working.
00:13:56saratogacomparing to this page for the H10 might be interesting: http://www.rockbox.org/wiki/IriverH10PortDevInfo
00:13:59thomasjfoxsoap: Can you open a normal user terminal?
00:14:00gevaertsAlso the mr100, the samsungs, and I think the vibe500
00:14:15soapthomasjfox, yes
00:14:33thomasjfoxTry if you have the "root" or "sudo gainroot" stuff installed. Then you don't need the password
00:14:37gevaertsWell, some of them use a different filename, but that's not very important
00:15:20wah_wah_69thanks saratoga! I should have found thank link myself.
00:15:29soapthomasjfox, "Enable RD mode to gain root privileges"
00:15:49Torneinstall the rootsh package from maemo.org
00:15:53Tornethen sudo gainroot
00:16:14Torneyou can't become root without installing something or turning on R&D mode with flasher, but that has other sideeffects
00:16:47saratogawah_wah_69: also this page: http://www.rockbox.org/wiki/SamsungYH92xPort
00:18:19saratogadoes it also have a .ROM file in the system folder? if so that may contain the recovery code if you flash a bad .mi4 file
00:18:24soapTorne, yea that's what google is telling me. I would have sworn I went through all this ages ago.
00:18:45 Quit TheSeven (Disconnected by services)
00:18:46 Join [7] [0] (~TheSeven@rockbox/developer/TheSeven)
00:20:13wah_wah_69saratoga, yes it has a file called pp5020.mi4 in the system directory
00:20:38thomasjfoxgevaerts: I found something out about the pulseaudio CPU usage
00:20:48Tim_ElliottI wish someone would make a port for the Insignia Pilot
00:21:03thomasjfoxThe protection for the speakers just kicks in at a certain global volume level
00:21:13gevaertsTim_Elliott: New Year's wishes should be made in December
00:21:19thomasjfoxSo CPU is about 7% lower if I keep the volume below that
00:21:22AlexPNot on demand.. blah.. interested owners.. blah.. etc. etc.
00:21:31gevaertsthomasjfox: interestin
00:21:31 Quit feisar- (Ping timeout: 276 seconds)
00:21:33gevaertsg
00:21:37wah_wah_69it boots as USB mass storage just by plugging an usb cable, so it might be easy to recover
00:21:43 Quit bertrik (Quit: CGI:IRC)
00:22:43thomasjfoxgevaerts: Maybe we should set an initial volume on application startup just a bit below that level
00:23:41 Quit kugel (Remote host closed the connection)
00:23:44***Saving seen data "./dancer.seen"
00:23:47AlexPthomasjfox: can you detect speakers vs headphones, and then set a limit in rockbox a fraction below that when on speakers?
00:24:03AlexPAs it'd make no difference to the max, but save a load of power by the sounds of it
00:24:13thomasjfoxAlexp: Should be possible
00:24:26soapthomasjfox, big issue is that rockbox doesn't show up as a running application,
00:24:47gevaertsthomasjfox: do we want to touch the global volume at all?
00:25:02thomasjfoxgevaerts: Maybe just once?
00:25:18thomasjfoxsoap: But you are able to run it?
00:25:23 Quit tuxifier (Remote host closed the connection)
00:25:23gevaertsI know I'd be annoyed if apps all think they should set the volume for me
00:25:24wah_wah_69saratoga, it has some other files but none are called *.rom, there's a sysinfo.ini file which contains some info, is seems to have been generated by the firmware updater, this line is the one that references the bootloader: Bootloader=BL_pp6005_5020_color.rom
00:25:53thomasjfoxgevaerts: We could make it a platform based option
00:26:14soapthomasjfox, I have no music on my N810, but elsewise it runs fine. I'll load some music when I find a micro cable
00:26:35gevaertsthomasjfox: the way I see it, rockbox is an *application* on maemo. It shouldn't try to take over the system in any way
00:26:41thomasjfoxsoap: hehe, you really haven't used it in a long time
00:27:06thomasjfoxgevaerts: Hmm. On the other hand is saves quite some CPU
00:27:12soapI use it every day, but it is an appliance to me.
00:27:18gevaertsYes, but it's not CPU *we* are wasting
00:27:45soapeverything I need is installed (fbreader, claws mail, ssh) and that's all I do.
00:28:19thomasjfoxgevaerts: What if we make this a platform option and only set the volume if 1. it's on speakers and 2. it's above the limit (=don't touch lower volumes)
00:28:24 Quit BlakeJohnson86 (Quit: Leaving.)
00:28:55gevaertsthomasjfox: well, to put it simply, if I installed an audio player that did that, I'd report it as a bug
00:29:08thomasjfoxgevaerts: It should be configurable of course
00:30:15thomasjfoxI don't like the idea much either - but it's a cheap way to save CPU :)
00:31:44 Quit Keripo (Quit: Leaving.)
00:31:45 Quit ender` (Quit: Information travels more surely to those with a lesser need to know.)
00:32:51 Quit Buschel (Ping timeout: 265 seconds)
00:37:41 Join factor [0] (~factor@75.108.68.114)
00:39:12 Quit chattr (Ping timeout: 240 seconds)
00:41:43 Join chattr [0] (~mike@244.87.189.72.cfl.res.rr.com)
00:42:56thomasjfoxsoap: Can you just download a mp3 of the net for testing? Here's a free track: http://simonv.com/music/mp3/SIMON_V-Stand_by_V.mp3
00:43:42thomasjfoxsoap: Would be nice to get some CPU load info when the display is on/off
00:45:52 Join JdGordon| [0] (~jonno@rockbox/developer/JdGordon)
00:49:49 Join froggyman [0] (~seth@unaffiliated/froggyman)
00:50:51 Join BlakeJohnson86 [0] (~bjohnson@c-24-118-162-123.hsd1.mn.comcast.net)
00:51:39 Quit GuySoft (Ping timeout: 246 seconds)
00:53:30 Quit thomasjfox (Remote host closed the connection)
00:53:38 Join Strife89 [0] (~Strife89@adsl-80-182-157.mcn.bellsouth.net)
00:54:58 Quit Tim_Elliott (Ping timeout: 272 seconds)
00:58:31 Join GuySoft [0] (~guysoft@bzq-109-64-46-137.red.bezeqint.net)
01:00
01:00:33 Join Keripo [0] (~Keripo@CPE0022b0d4bdb7-CM001a6680d4fe.cpe.net.cable.rogers.com)
01:04:16 Quit GuySoft (Ping timeout: 276 seconds)
01:05:01 Join eRivas [0] (~je@190.150.0.99)
01:05:50eRivasis it possible to have a volume icon that will display volume level only when it is changed?
01:07:05 Join mikroflops_ [0] (~yogurt@h-34-71.A238.priv.bahnhof.se)
01:07:06n1syes
01:08:02n1sat least i'm pretty sure it is, take a look ath the CustomWPS page for the details
01:08:37eRivasin the manual?
01:09:09n1sthe wiki is probably better but the manual has the wps tags too
01:10:00eRivasok, I've already taken a look at the tags, what I don't understand is how to handle the events, since the numeric display won't be always visible
01:10:12 Quit mikroflops (Ping timeout: 246 seconds)
01:10:37JdGordon|umm, you should be able to do it
01:10:41JdGordon|let me have a quick look
01:11:15 Part domonoky
01:12:18JdGordon|eRivas: you can do %?if(%bl, >, 90)<charged>
01:12:40JdGordon|90 is the % level which you want to define as "close enough to fully charged"
01:12:58JdGordon|oh woops... changed not charged
01:13:03JdGordon|no, you cant do that
01:13:16eRivasI saw something like that in the failsafe theme
01:13:31JdGordon|you can do it for volume, not battery
01:13:43eRivasyeah, volume I said
01:13:57JdGordon|wtf? i'm still asleep here :p
01:14:20eRivashahaha, so an if is the way to go?
01:14:39 Quit kadoban (Ping timeout: 240 seconds)
01:14:53JdGordon|no, there is a actual tag for volume changing
01:15:22JdGordon|%?mv<%pv> is what you want
01:15:45eRivasok, the wiki has more of this?
01:15:48 Join GuySoft [0] (guy@bzq-79-179-5-239.red.bezeqint.net)
01:16:29JdGordon|http://www.rockbox.org/wiki/Main/CustomWPS#Increasing_47Decreasing_Volume
01:16:58eRivasexcellent, thanks a lot
01:17:13 Join feisar- [0] (jljhook@irkki.fi)
01:18:09eRivasso, if %mv is true, show numbers, else, show icon?
01:19:29JdGordon|that gets more complicated
01:19:30 Quit antil33t (Read error: Connection reset by peer)
01:19:58JdGordon|%?mv<%pv|%pv(stuff to make it a bar)> or you can do a icon strip also
01:21:30eRivasyes, I already have the strip
01:22:49eRivascurrent code:
01:22:50eRivas%?pv<%xd(da)|%xd(db)|%xd(db)|%xd(dc)|%xd(dc)|%xd(dd)|%xd(dd)>
01:22:52 Join Rob2222 [0] (~Miranda@p4FFF03A1.dip.t-dialin.net)
01:23:01eRivaswith volume icon declared as d
01:25:08 Join webguest91 [0] (~48802c70@giant.haxx.se)
01:27:25 Quit webguest91 (Client Quit)
01:27:49 Join webguest94 [0] (~48802c70@giant.haxx.se)
01:27:54 Quit webguest94 (Client Quit)
01:35:23 Quit n1s (Quit: Lmnar)
01:37:52 Join markun [0] (~markun@5ED33C2C.cm-7-4a.dynamic.ziggo.nl)
01:37:52 Quit markun (Changing host)
01:37:52 Join markun [0] (~markun@rockbox/developer/markun)
01:46:20eRivaswhen I declare and image to be preloaded, I can specify coordinates, are these coordinates relative to the viewport containing the image or are they absolute?
01:46:30 Join Keripo1 [0] (~Keripo@CPE0022b0d4bdb7-CM001a6680d4fe.cpe.net.cable.rogers.com)
01:48:26 Quit Keripo (Ping timeout: 250 seconds)
01:50:46 Join DerPapst [0] (~Alexander@91-66-226-46-dynip.superkabel.de)
02:00
02:01:22JdGordon|eRivas: viewport
02:02:43eRivasand can I use math operations, I'm thinking of displaying vol as positive integers, as opposed to the true -dB approach
02:02:53 Quit feisar- (Read error: Operation timed out)
02:03:18eRivasso if min vol is -81 dB, I would like to do %pv+81
02:04:23JdGordon|no
02:05:03 Quit mortalscan (Ping timeout: 240 seconds)
02:06:13 Quit froggyman (Ping timeout: 260 seconds)
02:08:40 Join AndyI [0] (~pasha_int@212.14.211.84)
02:10:20 Join antil33t [0] (~Mudkips@124-197-51-80.callplus.net.nz)
02:11:16 Quit antil33t (Client Quit)
02:11:38 Join antil33t [0] (~Mudkips@124-197-51-80.callplus.net.nz)
02:15:32 Join JesusFreak316_ [0] (~JesusFrea@pool-173-65-105-252.tampfl.fios.verizon.net)
02:23:47***Saving seen data "./dancer.seen"
02:26:17[7]oh my.
02:26:41[7]do i understand correctly that the ATA layer doesn't handle splitting too-large transfers for non-lba48 drives?
02:28:06 Quit JesusFreak316_ (Remote host closed the connection)
02:42:28*[7] spots a rockbox main menu on his ipod classic's LCD :)
02:42:48gevaerts\☺/
02:42:56gevaertsCongratulations!
02:43:05*[7] needs to figure out why the clickwheel doesn't work :/
02:43:30[7]might be as trivial as me forgetting to unmask an IRQ
02:45:11*[7] was right :)
02:45:31*[7] browses folders on the disk
02:46:51[7]four more things that need to be looked into rather soon:
02:47:00[7]- enable caches
02:47:10[7]- make codecs/plugins compile
02:47:20[7]- implement ata dma
02:47:37[7]- figure out why Buschel's LCD code doesn't work
02:59:20 Join kadoban [0] (~kadoban@ip98-165-177-158.ph.ph.cox.net)
03:00
03:00:57 Join feisar- [0] (jljhook@irkki.fi)
03:02:35 Join einarin [0] (~johnny@207-224-41-189.hlrn.qwest.net)
03:05:22 Quit feisar- (Ping timeout: 246 seconds)
03:05:37einarinis there someone around who could help me with the Sansa AMS recovery mode?
03:11:28saratogaeinarin: maybe, but i've never tried it myself
03:14:49 Quit timccc (Ping timeout: 246 seconds)
03:16:06eRivasI would like to suggest new strings available for translation: in, of
03:16:18 Join jduley [0] (~ca24b342@giant.haxx.se)
03:17:26 Quit DerPapst (Quit: Leaving.)
03:19:01*[7] spots a nice cabbiev2 backdrop :)
03:19:36[7]plugins fail with "incompatible model", whatever that means
03:20:06 Join mortalscan [0] (~mortalsca@173-166-0-166-newengland.hfc.comcastbusiness.net)
03:20:46saratogathey check to make sure they're compiled for the right target ID IIRC
03:21:00[7]yeah, but why doesn't it recognize it's own id?
03:21:07 Join timccc [0] (~timccc@112.166.15.141)
03:21:11[7]i.e. where did i forget to change it?
03:21:32saratogai think you have to set it in configure and maybe the target.h file
03:21:35saratoganot sure though
03:23:23jduleysaratoga: Hi, has anyone looked at fixing pause/unpause squeak on the clip+ FS #11812?
03:23:32saratogano i don't think so
03:23:38saratogai just noticed it recently actually
03:23:55saratogait doesn't happen with my usual headphones so i'm not too concerned ;)
03:24:39 Quit mortalscan (Ping timeout: 240 seconds)
03:25:11jduleyI just heard from another person who experiences it so was wondering if it was something worth fixing
03:26:30saratogaits worth fixing
03:26:44saratogaif someone posts a patch i'll confirm and then commit it
03:30:53jduleyok I wouldn't have a clue where to start with a patch so it won't be me. Do you think it happens on all AMSv2 devices as that other guy had it on his Fuze v2?
03:31:13soapThomasjfox, I promise to be less apathetic towards Rockbox on the N810 tomorrow if you highlight me.
03:32:56[7]hm, the target id in the plugins seems to be correct
03:33:03[7]where's the target id that the loader checks?
03:36:13saratogayou should probably ask kugel or someone in the morning
03:36:19saratogaplugins aren't the most important thing anyway
03:36:34[7]but codecs are :)
03:36:43saratogahmm yes good point
03:37:10*[7] wonders if it will reach 10% realtime in its current state
03:38:10saratoga[7]: codecs.c
03:38:12saratogaaround line 200
03:38:49[7]i spotted it in the plugin code in the meantime, and both values are 0x47, so it must be something else
03:39:01saratogaoh you said model, not version
03:39:04saratogaheh i have no idea
03:39:10saratogai'm actually a little surprised we check for that
03:41:03saratoga[7]: it compares TARGET_ID as defined in teh codec to TARGET_ID as defined in the main binary
03:41:47[7]it's probably the load address that's making it fail
03:41:58saratogayeah i don't see anyway they could be different
03:49:18CIA-7New commit by 03saratoga (r28942): Commit part of FS #11748 by Michael Hohmuth. Adds support for automatically resuming any song that is not played to completion at any point later in ...
03:50:02 Join froggyman [0] (~seth@unaffiliated/froggyman)
03:50:07[7]yeah, was indeed a linker script problem
03:50:29einarinsaratoga: on the sansaAMSunbrick page, it says to dd orig_image.bin to the device, but they don't say what orig_image is or how to make one
03:51:56saratogaeinarin: its just a copy of the front of the NAND chip from a working player
03:51:56CIA-7r28942 build result: 46 errors, 0 warnings (saratoga committed)
03:51:59saratogabah
03:52:21[7]hm, i'd really like to get that port into SVN, but it requires those ATA changes that haven't been tested on other targets
03:52:34saratogadamn archos!
03:54:39 Quit froggyman (Ping timeout: 240 seconds)
03:57:05saratogahuh i don't understand why archos is different here
03:57:24eRivashow do I use the %Re tag, just insert it or specify alternatives?
03:58:13 Quit GuySoft (Ping timeout: 260 seconds)
03:59:12 Join GuySoft [0] (guy@bzq-79-179-5-239.red.bezeqint.net)
04:00
04:00:26 Quit xavieran (Ping timeout: 255 seconds)
04:01:04 Quit [7] (Disconnected by services)
04:01:05 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
04:01:51 Join feisar- [0] (jljhook@irkki.fi)
04:05:57TheSevenwoah!
04:06:05TheSevenwe're getting closer
04:06:14*TheSeven needs to fix audio tomorrow
04:06:38TheSevenit plays back the first second correctly, and then it goes nuts
04:06:44saratogaok i have no idea whats going on, how can an extern variable in tagcache.c to playback.c work on SWCODEC but not on HWCODEC?
04:06:54 Quit feisar- (Ping timeout: 265 seconds)
04:07:05TheSevenUSB seems to work apart from it exposing a wrong sector size
04:08:46saratogawait
04:09:01saratogaHWCODEC uses playback.c right?
04:10:00*TheSeven falls asleep... it's 4AM again
04:13:25 Join xavieran [0] (~xavieran@ppp118-209-79-106.lns20.mel4.internode.on.net)
04:13:30saratogaugh this is going to be hard to fix without talking to a HWCODEC person
04:15:17JdGordon|saratoga: wassup?
04:15:39saratogai didn't realize HWCODEC works completely different then SWCODEC for handling track changes
04:16:08saratogai can make it compile, but i have no idea if this automatic resume feature will even work as implemented on HWCODEC
04:16:15saratogaand i have no device to try it on
04:16:40JdGordon|ahem... we should fork hwcodec out and drop it
04:16:44saratogayes
04:16:48saratogabut anyway
04:17:10saratogahow limited is HWCODEC? i thought it just used a DSP but was otherwise pretty similar in terms of buffering and playback, but it looks like that is not the case
04:17:25saratogadoes it make sense to support resume on it?
04:18:25JdGordon|I assume so
04:19:11saratogaanyone got a HWCODEC player handy to test something?
04:19:21JdGordon|Massive assumption, but I dont tihnk the hwcodecs would sell very well if they couldnt resume mid track :p
04:19:46saratogaor can I do this in the sim (does the sim even work work for HWCODEC)?
04:19:51 Quit jduley (Quit: CGI:IRC)
04:19:57JdGordon|sim is not simulated for hwcodec at all
04:20:05JdGordon|s/sim/playback/
04:20:06saratogaamazing
04:20:13 Join Barahir_ [0] (~jonathan@frnk-590f4df5.pool.mediaWays.net)
04:20:25saratogawho actually has one of these devices
04:20:51JdGordon|RockboxTesting ...
04:21:15JdGordon|I have an ajbr but not with me, and most of the hwcodec owners are *difficult* to get to test things
04:21:29saratogayes the wiki suggests you
04:21:41 Quit Barahir (Read error: Operation timed out)
04:21:58eRivascan I use playlist name - %pn - as a condition for an if?
04:22:06saratogaLlorean LambdaCalculus379
04:22:09saratogaone of you be online
04:22:14JdGordon|eRivas: yes
04:22:19eRivasok
04:22:30JdGordon|saratoga: #ifdef if out.. doesnt it require the db anyway which isnt on the lowmem targets?
04:22:34JdGordon|or am i confusing things?
04:23:04saratogathe player has the database
04:23:42JdGordon|eRivas: what sort of check do you want to do with it? I *think* %?pn is true when you play from a m3u (and maybe directory) and false when its a dynamic playlist
04:23:48***Saving seen data "./dancer.seen"
04:24:24eRivasJdGordon|: %?pn<%pn>
04:24:33JdGordon|yeah that would prob work
04:24:39eRivasthanks
04:26:38JdGordon|saratoga: two options, 1) #ifdef it out and leave it untill someone can test, 2) make it compile and wait for complaints/verification that it works
04:26:51eRivasJdGordon|: how do I use the %Re tag, just insert it or specify alternatives?
04:27:00saratogaif i understand mpeg.c correctly, its not going to work as currently implemented
04:27:05JdGordon|the danger with 2) is what happened to me, 6+months before finding out it is broken
04:28:33saratogawhat the hell does hwcodec actually include then
04:28:44JdGordon|eRivas: like all tags, it depends what you want... %Re will show some text, %?Re<option|option|option> gives you more control
04:28:45saratogai just realized i don't know if i've ever looked at a rockbox source file it uses before now
04:29:23eRivasJdGordon|: it's just that I insterted it alone and will only show aiff, other encoders will not be shown
04:29:35eRivasJdGordon|: so I guess I must specify my text
04:31:51 Quit kadoban (Quit: bye)
04:32:07 Join kadoban [0] (~kadoban@ip98-165-177-158.ph.ph.cox.net)
04:39:36 Join fyrestorm [0] (~nnscript@cpe-69-203-144-35.si.res.rr.com)
04:39:46 Quit kadoban (Ping timeout: 265 seconds)
04:43:16 Quit GuySoft (Ping timeout: 264 seconds)
04:44:13 Quit JdGordon| (Ping timeout: 260 seconds)
04:49:23 Quit pixelma (Disconnected by services)
04:49:24CIA-7New commit by 03saratoga (r28943): Blind commit a 'fix' for automatic resume on HWCODEC since I don't understand HWCODEC and have no way to test builds for it. For now just disable it. ...
04:49:26 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:49:28 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:50:50 Quit amiconn (Disconnected by services)
04:50:51 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:50:54CIA-7r28943 build result: All green
04:51:08 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:51:10saratogafuck still wastes 100 bytes on HWCODEC
04:51:38saratogaamiconn: (for the logs) can you comment on how this should be fixed in FS #11748?
05:00
05:01:31 Quit factor (Ping timeout: 265 seconds)
05:03:05 Join feisar- [0] (jljhook@irkki.fi)
05:03:10 Quit TheSeven (Ping timeout: 240 seconds)
05:07:43 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
05:14:37eRivasis there a radio token to indicate radio muted?
05:14:59 Join factor [0] (~factor@75.108.68.114)
05:16:41einarinsaratoga: thanks
05:17:06 Quit Rob2222 (Ping timeout: 272 seconds)
05:17:42 Quit wah_wah_69 (Ping timeout: 276 seconds)
05:49:42 Quit einarin (Ping timeout: 276 seconds)
05:51:00 Join wah_wah_69 [0] (~io@212.225.156.123)
05:51:18 Join einarin [0] (~johnny@207-224-41-189.hlrn.qwest.net)
05:57:33 Quit MethoS- (Remote host closed the connection)
06:00
06:01:06 Join fyre^OS [0] (~nnscript@cpe-69-203-144-35.si.res.rr.com)
06:03:34 Quit eRivas (Quit: Saliendo)
06:04:52 Quit fyrestorm (Ping timeout: 264 seconds)
06:07:50 Join JdGordon| [0] (~jonno@124-149-188-197.dyn.iinet.net.au)
06:07:50 Quit JdGordon| (Changing host)
06:07:50 Join JdGordon| [0] (~jonno@rockbox/developer/JdGordon)
06:09:13 Quit einarin (Quit: leaving)
06:11:42 Quit wah_wah_69 (Read error: Connection reset by peer)
06:14:08 Quit parafin (Ping timeout: 276 seconds)
06:20:15*JdGordon| wonders if hwcodec can be simulated on swcodec by setting up a virtual MAS
06:23:36 Join binaryhermit [0] (~binaryher@adsl-70-131-127-225.dsl.emhril.sbcglobal.net)
06:23:52***Saving seen data "./dancer.seen"
06:24:34 Join kadoban [0] (~kadoban@ip98-165-177-158.ph.ph.cox.net)
06:26:00 Join binaryhermit_ [0] (~binaryher@adsl-99-135-145-17.dsl.emhril.sbcglobal.net)
06:26:13 Quit binaryhermit_ (Read error: Connection reset by peer)
06:28:35 Quit factor (Ping timeout: 240 seconds)
06:30:04 Quit binaryhermit (Ping timeout: 264 seconds)
06:48:38 Quit JdGordon| (Quit: Lost terminal)
06:49:29 Join JdGordon| [0] (~jonno@124-149-188-197.dyn.iinet.net.au)
06:49:29 Quit JdGordon| (Changing host)
06:49:29 Join JdGordon| [0] (~jonno@rockbox/developer/JdGordon)
07:00
07:09:28 Join GuySoft [0] (guy@bzq-109-67-22-249.red.bezeqint.net)
07:16:12 Quit Keripo1 (Quit: Leaving.)
07:45:33 Join leavittx__ [0] (~lev@MS-159-112.dyn-ip.SPb.SkyLink.RU)
07:46:39 Join Thopter [0] (~AbuMaia@208.123.144.34)
07:49:10 Quit leavittx_ (Ping timeout: 276 seconds)
07:50:42ThopterI'm new to Rockbox, having installed it via the Rockbox Utility on my Clip+ 8GB earlier today. I found a language quiz patch that I would like to install, but I'm not sure how to go about doing that. All the info I've read about adding patches talks about using diff files, but this patch is provided as a .c file. Can someone point me to some instructions somewhere that could get me through?
07:52:48 Join Horschti [0] (~Horschti@xbmc/user/horscht)
07:57:07 Quit Horscht (Ping timeout: 276 seconds)
08:00
08:01:01 Quit xavieran (Ping timeout: 276 seconds)
08:05:00 Join froggyman [0] (~seth@unaffiliated/froggyman)
08:11:49 Join parafin [0] (parafin@paraf.in)
08:13:03 Quit JdGordon| (Quit: Lost terminal)
08:13:49 Join xavieran [0] (~xavieran@ppp118-209-79-106.lns20.mel4.internode.on.net)
08:23:54***Saving seen data "./dancer.seen"
08:26:14 Join saratoga_ [0] (9803c22e@gateway/web/freenode/ip.152.3.194.46)
08:31:53 Join mortalscan [0] (~mortalsca@173-166-0-166-newengland.hfc.comcastbusiness.net)
08:39:09 Quit user890104 (Ping timeout: 272 seconds)
08:40:04 Join user890104 [0] (~Venci@6bez10.info)
08:48:41 Quit antil33t ()
08:53:00 Join JdGordon [0] (7c95bcc5@gateway/web/freenode/ip.124.149.188.197)
09:00
09:08:30 Quit fyre^OS (Quit: Ur skills' fireproof like a wooden panel -- U got feds talking leet on your IRC channel!)
09:11:29 Join [Saint] [0] (S_a_i_n_t@203.184.3.36)
09:36:29 Quit JdGordon (Ping timeout: 265 seconds)
09:46:23 Join bmbl [0] (~bmbl@unaffiliated/bmbl)
10:00
10:00:25 Quit xavieran (Ping timeout: 246 seconds)
10:14:36 Join xavieran [0] (~xavieran@ppp118-209-79-106.lns20.mel4.internode.on.net)
10:16:44amiconnTheSeven: DMA for LCD updates still has advantages even if the main thread needs to be blocked until completion: Other threads can run meanwhile (e.g. codec)
10:17:47 Quit mortalscan (Ping timeout: 240 seconds)
10:23:15 Quit BHSPitMonkey (Remote host closed the connection)
10:23:57***Saving seen data "./dancer.seen"
10:33:31amiconnsaratoga: Hwcodec does not use playback.c at all. It has its own (considerably less complex) playback engine, in mpeg.c
10:34:12 Join icarusfactor [0] (~factor@75.108.68.114)
10:35:14 Quit icarusfactor (Remote host closed the connection)
10:35:37 Join icarusfactor [0] (~factor@75.108.68.114)
10:38:12 Join factor_ [0] (~factor@75.108.68.114)
10:38:17 Join n1s [0] (~n1s@rockbox/developer/n1s)
10:38:57*pixelma doesn't think she is "difficult to test things"
10:39:04pixelma+get to
10:39:16 Nick icarusfactor is now known as factor (~factor@75.108.68.114)
10:39:40pixelmathe only thing about testing on the Ondio is that putting on test builds takes ages (if I need everything)
10:39:53 Quit factor_ (Read error: Connection reset by peer)
10:42:45 Join ender` [0] (krneki@foo.eternallybored.org)
10:53:30[Saint]It's amazing enough you even *have* the Ondio
10:53:38[Saint]let alone more than one.
10:54:38pixelmahuh, I only have one Ondio. And it's a nice player with better sound quality than the other players I have
10:55:17[Saint]Oh...sorry. I thought you had two. And yeah, they're not a bad player at all. I've only ever seen one "in the wild" though.
11:00
11:00:05saratoga_amiconn: so basically if we want to have this work, I'd have to figure out how mpeg.c works then tie it into that playback engine at the right spots?
11:03:21pixelmasaratoga_: does this feature need the database in RAM or only the database?
11:04:06saratoga_its about the automatic resume feature
11:04:30pixelmayes, but your commit message mentions that it is tied to the database
11:04:43saratoga_it requires some changes to playback.c (basically to actually set the resume positions)
11:05:35 Join stoffel [0] (~quassel@p57B4B791.dip.t-dialin.net)
11:05:36saratoga_it stores stuff in the database yeah, but the problem is actually getting track changes to use that information on HWCODEC
11:06:47pixelmait's just that if it needs the database in RAM then it's not worth looking into integrating it into hwcodec since those targets are lowmem so don't have the option to load the database into RAM (just that part isn't available there), that's al
11:06:49pixelmal
11:07:10saratoga_i don't think it needs to be in RAM
11:09:39pixelmaI remember that gather runtime data needed a few fixes to get to work properly on hwcodec but made to work in the end
11:16:22 Join fml [0] (~5df3c93d@giant.haxx.se)
11:16:50 Join pamaury [0] (~quassel@dhcp-129-228.residence.ens-lyon.fr)
11:16:50 Quit pamaury (Changing host)
11:16:50 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
11:16:56fmlHello. There is a typo in the last commit. In settings.h, the comment for the new settingis wrong (copy-paste?).
11:17:26fml...but I can't fix it now.
11:17:31 Quit fml (Client Quit)
11:17:44amiconnImo multi-resume is a rather esoteric feature. I fail to see a use case...
11:18:40pixelmasomething with podcasts and/or audiobooks
11:18:55amiconnFor automatic multi-resume, that is. Doing it manually has been possible for years already, using bookmarks
11:21:00amiconnBookmarks have the further advantage that they don't require the database, while the new multi-resume seems to
11:24:13[Saint]unfortunate;y, this conversation has been had before...I would have preferred to see functionality of bookmarking increased if it actually needed it.
11:24:21[Saint]*unfortunately
11:25:36 Join kevku [0] (~kevku@2001:7d0:0:f000::135d)
11:27:40pixelmaindeed, and I didn't really see a consensus or at least something close in the thread
11:27:58pixelmaI was even a bit surprised to see the commit now
11:29:01[Saint]As was I, I thought it was still very much "up in the air".
11:29:38Lloreanamiconn: The problem is the current bookmarking feature doesn't work with the database.
11:30:06LloreanTo me, the reasonable solution would've been to fix bookmarking to work with the database, and then add a "if a bookmark is present when file is played, automatically resume the most recent bookmark" option.
11:31:16[Saint]the impression I got was "yeah, but...that's too hard, and I already did this".
11:31:36pixelmaI don't complain too loud becaus I didn't take part of the discussion on the ml but I felt that all I wanted to say has been said (I mostly agree/d with Llorean) and the thread had already become so long :\
11:31:47pixelma*part in
11:32:37Llorean[Saint]: I don't like the idea that a less ideal solution should be used because a good one hasn't completed yet - they're hard to remove later, and it just leads to a pile of half-features interacting in weird ways.
11:32:39 Join stripwax [0] (~Miranda@87-194-34-169.bethere.co.uk)
11:32:57LloreanBut I think I'm in the minority of preferring to have things wait until a "good" solution to a problem comes along.
11:33:13*Llorean still feels Rockbox has gotten progressively buggy for a while now.
11:33:18pixelmaI think the same way
11:34:21[Saint]I've sat back, because the last time I got truly empassioned about something I thought was totally the wrong direction to head in...it fel on deaf ears.
11:34:31[Saint]As lons as it has an off button, fine with me.
11:34:46[Saint]*long
11:35:48 Join Topy44 [0] (~Topy44@f048144099.adsl.alicedsl.de)
11:36:29 Quit Topy (Read error: Connection reset by peer)
11:37:42 Join JdGordon| [0] (~jonno@124-149-188-197.dyn.iinet.net.au)
11:37:42 Quit JdGordon| (Changing host)
11:37:42 Join JdGordon| [0] (~jonno@rockbox/developer/JdGordon)
11:38:11saratoga_imo the really buggy stuff is in playback.c, and this doesn't seriously impact that one way
11:38:39saratoga_since the changes are pretty trivial and don't impact how playback.c works unless you enable automatic resume
11:39:08pixelmabut the point of "implemented half-baked feature" still stands, IMO
11:39:08saratoga_otherwise you just 'else' into the old case
11:39:17saratoga_i don't think so
11:39:24saratoga_its not obvious to me how you could implement this better
11:44:12LloreanFix bookmarking instead of adding a new feature? It serves basically the same purpose.
11:47:06saratoga_from my point of view bookmarking is essentially an even worse implementation of a less useful feature
11:47:36saratoga_while having a manual way to set resume points could be useful, its a pretty silly way to do things when you can quite easily do it automatically
11:48:08saratoga_in the long term i just assumed we'd get rid of book marking and allow a way to manually set resume points in the database if people thought it was useful enough
11:48:33pixelmawhy is it less useful? And from a user's point of view, I prefer the bookmarking implementation since it doesn't need the database
11:48:53LloreanYou'd be losing a lot of functionality if you took away having multiple bookmarks in a single file.
11:49:05pixelmaI don't know what you mean with automatic and manual there
11:50:10[Saint]is there not already automatic bookmarking?
11:50:16saratoga_you prefer it because it doesn't do things the correct way?
11:50:18[Saint]didn't Torne work on that?
11:50:33Llorean[Saint]: Yes. But there's not automatic resume of bookmarks (unless I missed that being added) and bookmarks don't work with the database.
11:50:39saratoga_from a users point of view how is that a preference?
11:51:07pixelmasaratoga_: what is the "correct" way? Not relying on another feature I need to explicitely enable if I don't use it?
11:51:32saratoga_the database is the correct way to store metadata about tracks
11:52:00saratoga_its literally what a database does
11:52:35pixelmaresume points aren't metadata in my eyes, it's a playback/playlist thing
11:53:13saratoga_playback data about a specific file thats unrelated to the current state of the playback engine is metadata
11:54:16LloreanPlayback state isn't really metadata in my eyes either.
11:55:09LloreanAlso the database is for storing it for search purposes. It's not like we're going to say "we're removing ID3 tags, files should store all their info in the database" or "we're removing support for external album art, it needs to be scanned and entered into the database"
11:55:17LloreanThe only functionality that should require the database is searching/using the database.
11:55:28saratoga_no
11:55:35saratoga_thats stupid
11:55:51saratoga_anything that involves storing data persistantly should use the database
11:55:59LloreanWhy?
11:56:04saratoga_because thats what its for
11:56:11LloreanNot really.
11:56:27LloreanIt was originally "tagcache" and I think the name gives a pretty clear reason for its being in existence.
11:56:27 Quit stripwax (Quit: http://miranda-im.org)
11:56:31 Quit bluebrother (Disconnected by services)
11:56:32 Join bluebroth3r [0] (~dom@rockbox/developer/bluebrother)
11:56:49saratoga_so what?
11:57:11LloreanSo, my reason's just as good as yours. You've decided its intent is "all persistent data" but that wasn't, at least, its original intent.
11:57:24LloreanAnd I don't think there's been a discussion that its intent has changed, has there?
11:57:37pixelmawhy is a resume point persistent? And I also think about bookmarking as related to the playlist and not to a certain file that happens to be affected
11:57:44saratoga_first of all, the database already stores more then just tags
11:58:16Lloreansaratoga_: What, file ratings?
11:58:27saratoga_run time stats as well
11:58:30AlexPI would fight tooth and nail against removing bookmarking and forcing the database
11:58:42LloreanRatings are something that traditionally goes into tags, only we don't support writing to tags.
11:58:48AlexPI don't object to this feature, but do not force the database on us
11:59:05LloreanRun time stats are necessary for database searches since that's something you can query it for.
11:59:17saratoga_like, say resume information?
11:59:46LloreanYou can query for resume information? What, are you going to search for "files I've got resume points in at longer than 4 minutes in"?
11:59:54AlexPbookmarks can e.g. be easily copied between players
12:00
12:00:00saratoga_"i've got a file, does it have a resume point?"
12:00:19Llorean"I've got a file, does it have album art"
12:00:26LloreanClearly album art needs to be using the database only too, now
12:00:44saratoga_no because we have to buffer album art
12:01:26saratoga_look i'm not doubting that you can duplicate everything that a sane data storage system does with a mess of files on the hard disk, i'm just saying doing that for no reason is ridiculous
12:01:49LloreanResume points without database isn't "no reason"
12:01:55AlexPno reason?
12:02:02saratoga_there are obviously a few targets where memory is so tight we can't do things sensibly, but they're fewer and fewer every day
12:02:08LloreanResume point are something people have been using, as bookmarks, without the database for *years* in Rockbox.
12:02:18LloreanThere's no reason to add it to the database other than a desire to force users to lose existing functionality
12:02:24saratoga_"without the database" is not a reason
12:02:31saratoga_literally
12:02:43saratoga_it doesn't explain anything
12:02:44LloreanWhat's the reason *for* adding it other than "I think it belongs there"?
12:02:55AlexPI don't really see the point in this discussion, it looks like it is going to get committed anyway
12:03:02saratoga_its already committed
12:03:06AlexPexactly
12:03:14AlexPSo why discuss if it is ignored
12:03:18LloreanAlexP: Yes, but now he's advocating removing bookmarks.
12:03:19 Quit factor (Ping timeout: 255 seconds)
12:03:27LloreanThat hasn't happened yet.
12:03:34AlexPLlorean: I'm not worried about that, I'd just revert it
12:03:36saratoga_i'm not removing bookmarks because i don't care
12:03:50saratoga_i'm just pointing how ridiculous people are about the database
12:03:58AlexPIn your opinion
12:03:59saratoga_avoiding it is an end in and of itself
12:04:10saratoga_not one person has so far mentioned why they don't want to use it
12:04:16Lloreansaratoga_: Bookmarking a single file shouldn't depend on having enough disk space for the whole database.
12:04:18saratoga_isn't that a little funny to you guys?
12:04:22AlexPLoss of lots of RAM
12:04:23LloreanA bookmark is tiny, the database isn't necessarily
12:04:31saratoga_how much RAM?
12:04:36AlexP2 MB
12:04:40LloreanIf there were a way to store it in the database without initializing it with tag data, sure, that's different.
12:04:46saratoga_2MB for how many tracks?
12:04:52AlexPLoads of small ones
12:05:12saratoga_how many?
12:05:43Lloreansaratoga_: forcing people to use the database because you think it's sane seems to be another end in and of its self.
12:05:45AlexPI can't remember exactly, my player isn't to hand. 15k ish IIRC although that is a guess
12:05:50LloreanWhy make them use it for a feature that doesn't *require* it?
12:06:00LloreanDatabase searches require it, obviously, by their nature. Bookmarks don't.
12:06:18saratoga_forcing people to use rockbox features that are implemented well is a goal of mine yes
12:06:26AlexPI can move bookmarks between players
12:06:29saratoga_you're complaining that i want to improve things
12:06:32AlexPHow to do that with the db
12:06:32saratoga_well played
12:06:37Lloreansaratoga_: That's not an improvement.
12:06:47saratoga_your technical opinion?
12:06:49Llorean"We'd like to force you to use an unnecessary amount of disk space for information that's a few bytes. "
12:07:07LloreanWhy should a user be force to cache all of their files' metadata just to store a resume point?
12:07:22LloreanThat's inefficiency at its fines.
12:07:24Lloreant
12:07:24saratoga_i'm not saying they should
12:07:26AlexPanyway, as I say I don't object to this particular feature too much, it is bookmarking I care about
12:07:34Lloreansaratoga_: You said they should be forced to use the database for resume points.
12:07:37saratoga_yes
12:07:38LloreanThat's what the database *is*
12:07:43saratoga_it doesn't have to be
12:07:58LloreanWell you didn't say you were talking about a hypothetical future database that follows rules known only to you.
12:08:06AlexPBeing able to have only those files with resume points added to the database would be very nice
12:08:08LloreanWe all assumed you were talking about the real database that exists right now.
12:08:21AlexPadd them one by one as resume points get added
12:08:23LloreanAs I said, I wouldn't object if the database could be restricted to only holding the resume data.
12:08:29saratoga_obviously i'm not since the database i'm talking about includes bookmarks . . .
12:08:44Lloreansaratoga_: Sorry, "the database we have plus the feature you explicitly mentioned adding to it"
12:08:58LloreanWe can't read your mind, so we obviously can't know about features you *haven't* spoken of.
12:09:08saratoga_well if we're talking about how we want things to be i don't see why we should assume limitations we have now
12:09:21saratoga_you could ask me
12:09:28AlexP!
12:09:48LloreanThat makes no sense. I shouldn't have to ask you about every possible permutation. If you're discussing one possibility you should describe it, not assume people can read your mind and know what you intend.
12:10:09saratoga_err, what?
12:10:10LloreanIf you say "I think the database should be the only place to keep resume points" it's pretty reasonable that someone will think you mean the *current* database, not some other one you've got in your head.
12:10:33 Join DerPapst [0] (~Alexander@91-66-226-46-dynip.superkabel.de)
12:10:39saratoga_no thats not reasonable since i'm talking about improving the database!
12:10:46LloreanEspecially when someone has already mentioned they'd be okay with it if it didn't have to include all the metadata, and you ignore that for five minutes without saying "yes, like that"
12:11:51saratoga_honestly no one even mentioned memory use until the very end of this argument, so i don't know why i should assume thats what you're concerned about
12:12:01Lloreansaratoga_: It's perfectly reasonable to think, when someone refers to the database, that they're referring to it as it exists, plus anything they've explicitly mentioned, because someone with the intent of having a constructive conversation would mention other changes they have. ESPECIALLY when they might mitigate some of the objections people brought up.
12:12:06 Join factor [0] (~factor@75.108.68.114)
12:12:58Lloreansaratoga_: (5:04:39 AM) Llorean: If there were a way to store it in the database without initializing it with tag data, sure, that's different. <−− Seems that even without the mention of "memory" this should've at least prompted a "they might be interested in this idea"
12:13:11saratoga_i think its perfectly reason for people who have a problem with a specific aspect of something (e.g. having a database in memory) to actually mention that instead of complaining about nothing in particular
12:13:32Lloreansaratoga_: You can't discuss a feature without being at the same place about it, so are there any other changes you're holding in your mind that might affect this image of the database you have?
12:14:20saratoga_yeah but i haven't hashed them out yet
12:14:54saratoga_i was thinking of a system where you either just index all the files but don't cache the tags, or better yet don't index anything until its actually played
12:15:13saratoga_so it behaves sort of like a cache on low memory targets
12:15:16LloreanHow about never index anything, if the user doesn't want to have it indexed?
12:15:43LloreanAs I said, I wouldn't mind if all my bookmarks or runtime stats were stored, but I don't want to waste disk space creating a searchable database I'll never use.
12:24:00***Saving seen data "./dancer.seen"
12:28:32 Join Buschel [0] (~chatzilla@p54A3B908.dip.t-dialin.net)
12:30:19Buschel[Saint]: ping
12:30:43[Saint]yo.
12:31:06 Join MethoS- [0] (~clemens@134.102.106.250)
12:31:14*TheSeven wonders if the iPod Classic port should go into SVN or FlySpray
12:31:34TheSeventhe problem is that it might break other targets that I don't have and thus can't test on
12:31:34 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
12:31:41TheSeven(due to the ATA changes)
12:31:44Buschelstill need verification for some patches on the iPod color. do you have access to it?
12:31:45[Saint]which ones?
12:32:08[Saint]Bushmills: Ah, yes. Which patches.
12:32:09BuschelTheSeven: if it possibly brakes other targets I would not submit to svn
12:32:48TheSevenBuschel: on the other hand, nobody seems to be willing to test that part right now
12:32:57TheSevenso this thing might be stuck in flyspray forever
12:33:31TheSevenif it breaks, it breaks completely (no disk access possible at all), so it will be apparent
12:33:36AlexPTheSeven: Send a mail to the dev list saying here it is on fs, please test as it might break x, y, z - if nobody tests it'll go in in a week
12:33:36TheSevenand probably trivial to fix
12:34:18AlexPI'm happy to test later on today
12:34:32*TheSeven needs to figure out which targets might be affected
12:34:54Buschel[Saint]: FS #11843 v16, and FS #11820 (will provide a test patch later)
12:34:54TheSevenbasically everything that's doing ATA byte swapping, and we should also test the bigendian targets
12:36:03BuschelTheSeven: why not submit the Classic port except this breaking change in one step and the ATA stuff in a second submit to be able to roll it back easily?
12:36:26TheSevenbecause the classic port can't work without the changes
12:36:31Buschel[Saint]: also, can you check which lcd_type your color has? this is available in the debug menu
12:37:15TheSevenand i don't think it will need a rollback. if anything, it will need a few more #ifdefs swapping bytes
12:37:28[Saint]Buschel: The OFs or RBs?
12:37:33BuschelRB's
12:37:33[Saint](debug menu)
12:38:17[Saint]LCD type: 1
12:38:31[Saint](for both, incedentally)
12:38:55Buschelhmm, interesting. both reach ~52 fps RGB full screen, right?
12:39:16[Saint]it was something like that.
12:39:33Buschelthe nano 1G also uses lcd_type 1
12:40:15[Saint]quite a difference in screen size though ;)
12:40:59Buschel[Saint]: yes, but the throughput in MB/s is still 2.5x faster nano1g vs. color
12:41:19Buschel[Saint]: short test regarding FS #11820 -> http://pastie.org/1423126
12:43:33Buschelamiconn: do you still have access to the iPod photo which you used to measure LCD speed (see LcdFrateRate wiki)
12:44:34Buschel[Saint]: if FS #11843 v16 works for you, I am interested in RGB/YUV speed boosted/unboosted. I would like to submit if it works fine
12:45:21[Saint]Ok, I'm patching now. Then building, <slight delay>, testing. :)
12:45:46Buschelso, you are able to build again :)
12:46:14[Saint]Yes, I figured out my CygWin gremlin.
12:51:15[Saint]rockboxdev.sh was trying to delete the build dir while it was still in it...the code for it was fine, I added a pause and prompt for manual resume and it worked fine.
12:52:01[Saint]without it, it was ignoring a "cd" for some reason, seems to work on other systems though.
12:52:11[Saint]*cd ..
12:58:21 Quit saratoga_ (Quit: Page closed)
13:00
13:01:13 Quit JdGordon| (Quit: Lost terminal)
13:04:07amiconnsaratoga: One main point about the database in rockbox is that it's optional. Atm I don't need bookmarks or automatic resume points, because I neither listen to audio books nor podcasts
13:05:36amiconnBut if I ever start doing that, I much prefer a way that doesn't require me to use the database, which I don't use regularly either
13:06:55amiconnI have it enabled for pictureflow to work, but that's only for occasional, show-off. In >99.9% of all cases I start playback from the file browser
13:08:02*[Saint] only has the DB enables so he can occasionally show someone pictureflow/wps integration.
13:08:23[Saint]*enabled
13:08:39*amiconn never even tried pf/ wps integration
13:09:29[Saint]kinda cool. But just for "eye candy".
13:09:52amiconnRockbox has a lot of feature I have no use for...
13:10:20amiconnAt least most of them aren't duplicating functionality.. this latest one does though :(
13:10:52[Saint]<y general theory is "as long as I can turn it off, then sweet"...this is a bit of a grey area.
13:11:34[Saint]It's only by the coincidence I use the DB for pictureflow that it doesn't really affect me. I understand there are many that don't use the DB.
13:13:41Buschel[Saint]: still building?
13:19:16 Join Judas_PhD [0] (~kevin@misterfluffy.dsl.xmission.com)
13:19:32 Quit leavittx__ (Ping timeout: 276 seconds)
13:19:36n1sspeaking of testing, could anyone with a coldfire target that's not h300 or hd300 test the testbuilds i linked in my mail to the ml last week? And it would be cool if someone could test building the new toolchain on cygwin too
13:19:58 Quit Judas_PhD (Client Quit)
13:21:56pixelmahmm, what do I need for the toolchain?
13:22:17pixelmado, I need to build it myself?
13:22:20pixelma-,
13:23:23pixelman1s: could you provide an M5 build with FM enabled (in advanced options)?
13:23:37n1spixelma: sure
13:24:56n1spixelma: it would be good if someone tested building the toolchain on cygwin
13:26:12pixelmaI'm not sure I know everything that's involved for doing so (except taking time... ;) )
13:28:16n1si have a patch so anyone who wants to test building just needs to apply it and run rockboxdev.sh
13:29:11n1s(the tools patch in FS #7832)
13:30:10n1spixelma: http://dl.dropbox.com/u/17484767/rockbox-m5-fm.zip
13:32:40n1sI'm mostly worried about target specific drivers breaking as everything seems to work fine on my h300
13:34:58pixelmaok, I'll have a look at sound etc. and recording and have an eye on the RTC. The radio is basically the same as the H300's AFAIK but you'll get half an X5 test with this too ;)
13:35:36n1sgreat
13:36:16 Join bluebrother [0] (~dom@g226069187.adsl.alicedsl.de)
13:36:16 Quit bluebrother (Changing host)
13:36:16 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
13:38:02DreamxtremeTheSeven i have a 3rd gen on the way if you need help to test anything
13:39:22 Quit bluebroth3r (Ping timeout: 272 seconds)
13:45:52 Join T44 [0] (~Topy44@f048048039.adsl.alicedsl.de)
13:48:44amiconnn1s: There's something in your 'rockboxdev.sh' patch that doesn't belong there
13:48:54amiconn'make' -> 'make -j4'
13:49:00 Quit Topy44 (Ping timeout: 240 seconds)
13:49:47 Join Ogham [0] (~Ogham@unaffiliated/ogham)
13:50:48OghamCan anyone throw any light on bug FS #11812 ? Does this have the potential to damage low impedance IEM's?
13:51:31*amiconn isn't sure which of the diff files in fs #7832 are needed to get the whole thing
13:53:34pixelman1s: playback works (flac and wav tested), sound settings seems to work (bass, treble tested which are in software with this DAC), voice works, FM seems to be working correctly, recording works (tested with the built-in mic), the RTC doesn't show any problems. Anything else to test?
13:54:40*amiconn could test h100 if he knew what's needed
13:55:25pixelman1s: oh - buttons work, mpegplayer works
13:57:27amiconnBuschel: Sure, since it's mine. But I don't have it with me atm
14:00
14:15:29 Join efyx [0] (~efyx@lap34-1-82-225-185-146.fbx.proxad.net)
14:18:25n1samiconn: only the last 2, the codec tuning one isn't actually necessary though
14:18:46n1sso just the patch for configure and rockboxdev.sh
14:19:19Buschelamiconn: fine, could you just test (function, speed) FS #11843 v16 and check what lcd_type you have? you could place this info in flyspray
14:19:56n1spixelma: sounds good, if you have a remote, testing that might be interesting
14:20:13*Buschel wonders where [Saint] is, he seems to have vanished...
14:20:48pixelman1s: don't have one yet, though B4gder wanted to send me one sometime
14:20:51n1samiconn: since it seems to work well for all tested players so far, things specific for the h100 would be most interesting, and the remote if you have one, since i don't
14:21:07amiconnI do have a remote
14:21:40amiconnThe H100 and the M3 have some drivers which the later coldfire targets don't have
14:21:58n1sand, yes i forgot to remove that make -j4 change as well as the commented out code since i wasn't sure if we wanted to keep support for the old toolchain in the script so that patch isn't quite finished
14:22:32amiconnThese targets do not have the PCF5060x, but separate ADCs instead. Especially the M3 one is rather picky about timings (and slow...)
14:23:05n1sit would be great if you could test that
14:23:20n1spixelma: ok, thanks for testing :)
14:23:45pixelmayou're welcome :)
14:24:03***Saving seen data "./dancer.seen"
14:25:28pixelmaCould it be that mp3 decoding got a bit faster or somesuch? I have the impression that mpegplayer works better (my videos I wanted to encode again because of choppy playback due to a bit higher bitrate look better and I wondered why)
14:30:59n1smp3 decoding got about 3% faster but i don't think that would be noticable in mpegplayer
14:31:52*amiconn is trying to build the new toolchain on cygwin
14:32:29n1samiconn: cool :)
14:33:03amiconnWill take a while, even though I've disable the on-access AV for this
14:33:07amiconn*disabled
14:33:51pixelmamaybe for other reasons then, anyway - it looks like I don't have to encode these videos again :)
14:34:02n1spixelma: nice
14:34:33n1samiconn: do you remember why the old toolchain needs a patch to disable some multilib?
14:34:39amiconnyes
14:35:10amiconnThe disabled multilib fails to build on various systems. Those need the patch
14:36:09n1sdo you know if it is needed for the new toolchain? (i guess your build will fail if that's the case)
14:37:05amiconnHow should I know? There's just one way to find out...
14:37:50n1syeah
14:38:20pixelmaguess we also need some testers for the toolchain on Mac too, IIRC they also show weird behaviour sometimes
14:38:50n1sis any active dev using macOS ?
14:39:01amiconnn1s: Do you use linux x86_64?
14:39:06n1syes
14:39:18amiconnSo it's probably not needed anymore.
14:39:27pixelmaLambda has one I think (and JdGordon had access to one but I don't know if this is the case anymore)
14:39:37n1sno, the 64 bit patch isn't needed anymore
14:40:13 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
14:40:50amiconnAh, that was another patch
14:41:09n1syes
14:44:08n1soh, i just remembered that the patched rockboxdev.sh only builds coldfire multilibs
14:45:00n1sso that particualr problem will probably not show up
14:45:08[Saint]did something with the speex encoder change recently...ish?
14:45:22[Saint]I can't seem to build voice files on CygWin
14:46:08amiconnn1s: Maybe it would be a good idea anyway to disable all multilibs we don't use. It would speed up tollchain building, and the resulting toolchain would need less disk space.
14:46:48n1syes, that is true
14:47:22amiconnIt would mean additional work if we add a target that needs them though
14:47:25[Saint]http://pastebin.com/8bkEEdPS is what bash has to say about the whole voice building fiasco, fwiw.
14:47:57amiconnSame applies to the SH multilibs btw - we only use sh1
14:48:40n1samiconn: the −−with-arch=cf config option makes it build only the coldfire multilibs, not the other m68k ones so maybe that's good enough?
14:49:03amiconnaha
14:49:08amiconnLet's see
14:51:38n1sonly 8 different multilibs are built and they take about a half megabyte of space each
14:53:58n1sthe old toolchain built 10
14:55:28Buschel[Saint]: does FS #11843 work for you?
14:55:56 Join stsquad` [0] (~user@nat18.sesnet.co.uk)
14:56:21amiconnn1s:
14:56:22amiconnconfigure: error: in `/tmp/rbdev-build/build-binutils/ld':
14:56:22amiconnconfigure: error: C compiler cannot create executables
14:59:48n1shmm
15:00
15:00:14n1si have to go for a while but will look at that later
15:00:38amiconnHappened after running for 25 minutes
15:03:19 Quit [Saint] (Ping timeout: 255 seconds)
15:09:24 Quit sasquatch (Ping timeout: 240 seconds)
15:11:44 Join [Saint] [0] (S_a_i_n_t@203.184.0.102)
15:13:41[Saint]Buschel: very, very weird behaviour.
15:14:31[Saint]the top of the screen is being drawn again oer the bottom of the screen.
15:14:52[Saint]which dissappears momentarily whilst scrolling.
15:16:19BuschelYUV is working fine?
15:16:42[Saint]how am I to establish that?
15:16:59Buschelplay a movie or use test_fps
15:17:35[Saint]yes.
15:19:20[Saint]the boot splash and the menus are garbled in the same way, a split about two thirds down the screen where the top of the screen is drawn overtop.
15:19:44BuschelI am checking right now
15:22:05Buschelcan you backtrace which patch-version introduced this? you'll just need to re-build rockbox.ipod
15:22:50Buschelseems like the RGB screen update is broken
15:24:00 Quit MethoS- (Remote host closed the connection)
15:24:08Buschel(which is funny as it did not change as much as YUV)
15:25:56 Join sasquatch [0] (~username@46.114.82.65)
15:37:22 Part stsquad` ("ERC Version 5.3 (IRC client for Emacs)")
15:39:58 Quit GodEater (Read error: Operation timed out)
15:42:43*Buschel cannot recognize the fault in the code... :/
15:46:48n1samiconn: that error seems vague, is there anything interesting in the config.log ?
15:49:55n1sthat the binutils build fails is interesting since it's the same version that we use for arm-elf now
15:50:05n1seh, arm-eabi-elf
15:54:24 Join benedikt93 [0] (~benedikt9@unaffiliated/benedikt93)
15:55:06 Join kugel [0] (~kugel@rockbox/developer/kugel)
15:55:52 Quit kadoban (Read error: Operation timed out)
15:56:36Buschel[Saint]: are you still there?
15:58:23 Join mystica555_ [0] (~mike@c-75-70-179-25.hsd1.co.comcast.net)
16:00
16:00:07Buschel[Saint]: yeah, I think I have it :)
16:03:49Buschel[Saint]: as the full screen lcd update needs to loop twice to update your screen, I need to correct the address within the framebuffer after the first run. this is not needed for the nano1g.
16:08:51TheSevendamn, that DMA buffering logic is driving me nuts
16:09:43 Join GodEater [0] (~bibble@rockbox/staff/GodEater)
16:10:28Buschel[Saint]: updated FS #11843 v17
16:12:31Buschel[Saint], soap: if I am not wrong the swapping in the LCD driver is not really needed. I will provide a patch for testing as soon as v17 (or any neccessery bugfix for this) has been submitted
16:15:34n1shmm, i have no idea what could cause that cygwin failure, maybe i should set up a cygwin environment to test in myself :/
16:17:12n1sapparently a new binutils release has been made but i can't find any info on what they have fixed in it, only added features
16:24:06***Saving seen data "./dancer.seen"
16:28:19 Quit mystica555_ (Ping timeout: 264 seconds)
16:34:14 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
16:34:24 Quit Buschel (Ping timeout: 276 seconds)
16:35:06TheSevenurgh
16:35:14TheSeventhat bugger doesn't even manage MP3 realtime
16:36:05TheSevenIIUC it's an ARM9e running at 216MHz
16:36:18TheSevencaches should be enabled
16:36:34TheSevenfrom the skipping behavior i'd say 80% realtime
16:37:02n1sthat sounds too slow for that core/freq even with caches disabled
16:37:41TheSevenwith caches disabled the UI was lagging
16:38:26TheSevenwhat's the best way to figure out the real clock?
16:38:49TheSevenlet it iterate a thousand times through a loop of a thousand nops and measure the time?
16:39:39n1si can't think of another way of doing it without knowing how it's configured
16:40:30TheSeventhe rest of the UI is responding quite fast, and games like chopper run fluently
16:44:07TheSevenn1s: 10.000 iterations of 1.000 nops took 52.319s
16:44:45TheSevenso if there would be zero overhead, that would be ~191MHz
16:45:10TheSevenit must be something different...
16:45:32n1sif caches are disabled and memory is slow enough to cause mp3 to not be realtime i'd guess the overhead would be significant
16:46:11n1scan you run test_codec ?
16:46:47n1san do less demanding codecs play fine, like flac or just a wav for example?
16:47:37 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
16:53:32kugelTheSeven: 10M instructions in 52µs?
16:54:55kugelthat's a lot more than 191MHz
16:55:13TheSevenhuh?
16:55:32TheSeven52ms (i used . as a thousands separator)
16:55:43kugeloh
16:56:14TheSevenwell, this is assuming that the usec timer counts usecs of course :)
16:56:31kugelyou don't use . as thousands seperator in english :)
16:58:01TheSevenwell, but then it would have been 10 instructions in 52s
16:58:36kugeltrue
16:59:06TheSevenand the usec timer seems to be sane
17:00
17:00:01TheSeven5'229'780s for 1'000'000'000 instructions took like 5 seconds, so that's fine
17:00:07 Join Keripo [0] (~Keripo@CPE0022b0d4bdb7-CM001a6680d4fe.cpe.net.cable.rogers.com)
17:00:11 Join mortalscan [0] (~mortalsca@173-166-0-166-newengland.hfc.comcastbusiness.net)
17:00:29TheSeven(191.212632MHz)
17:01:23TheSevendoes rockbox do cpu time accounting for the individual threads?
17:01:44TheSevenmaybe there's just some CPU hog?
17:03:18*TheSeven compiles a build with test plugins
17:03:50n1sthe Os stacks (IIRC) debug screen shows someinfo about the threads but not the cpu time i think
17:04:35TheSevengevaerts: any idea why rockbox usb exposes 512 byte sectors on a drive with 4096 bytes per physical sector, formatted as a superfloppy fat32 file system with 4096 bytes per sector?
17:05:07TheSeventhe file system works fine in rockbox, so that ipod video hack layer must be functioning
17:19:09TheSevenn1s: flac plays fine
17:19:27TheSevenbut the playback progress bar is jumping like crazy
17:19:48TheSevenprobably due to a problem with pcm_get_bytes_waiting
17:19:48n1sweird
17:29:36pamauryTheSeven: does usb msc supports something else than 512 bytes per sectors ?
17:29:51pamauryI don't remember but I'm not sure
17:30:18TheSeventhe whole flac file (27MB) would fit easily into RAM, and there's also 27MB allocated in the buffering debug screen
17:30:26TheSeventhe pcm buffer is always around half full
17:30:47TheSevenpamaury: certainly
17:31:45TheSevenreal is always in the 50000-70000 range, usefl is 20000-40000
17:31:56TheSevenso we're indeed facing a problem on the disk side of things?
17:32:12TheSeventhe ATA driver is still PIO only for now
17:32:46n1sif the disk is the bottleneck wouldn't mp3 work better than flac?
17:33:30 Join liar [0] (~liar@188-22-211-66.adsl.highway.telekom.at)
17:33:56TheSevenif disk accesses are eating lots of CPU, the MP3 codec might push the bandwidth further down
17:34:34TheSevenand the scheduler efficiency might also suffer at a more balanced codec/disk load
17:34:41TheSevenfrom*
17:34:49n1syeah, but the mp3 file will be much smaller than flac, anyway, what numbers do you get from test_disk?
17:35:14TheSevenlet me check
17:35:43TheSevencreate: 6 files per second
17:36:07TheSevenopen: 10568 files, dirscan: 487371, delete: 5
17:36:31TheSevencreate/write/read 512,a: 16/16/783 kB/s
17:36:46TheSevencreate/write/read 512,u: 16/16/715 kB/s
17:37:06TheSevencreate/write/read 4096,a: 924/936/789 kB/s
17:37:20TheSevencreate/write/read 4096,: 935/947/781 kB/s
17:37:30TheSeventhat was 4096,u
17:37:45TheSevencreate/write/read 1048576,a: 961/979/812 kB/s
17:37:58TheSevencreate/write/read 1048576,u: 943/956/807 kB/s
17:38:05n1sthat looks like a definite bottleneck
17:38:22TheSevenhowever read performance seems to be ok, even for sub-sector accesses
17:39:32n1sright, shouldnt' be bad enough to cause stuttering for mp3
17:39:40 Quit liar (Read error: Operation timed out)
17:40:09TheSevenhm, maybe disk accesses are over-yielding?
17:40:44n1ssee http://www.rockbox.org/wiki/DiskSpeed for a speed comparison
17:41:15n1sTheSeven: iirc a thread should yield about once each tick or so if not sleeping
17:42:22TheSevenhm, disk accesses are certainly yielding way more
17:42:35TheSeventhere will be several calls to yield() after each access
17:45:13 Quit sasquatch (Ping timeout: 240 seconds)
17:45:41 Join sasquatch [0] (~username@2.210.213.240)
17:46:28*TheSeven needs to reformat the disk
17:53:02TheSevenhm, after i replaced some yields that aren't present on other devices with busy waiting the buffering screen results for flac are a tiny bit better, but still way off normal
17:53:24TheSevenmp3 is still stuttering
17:55:37kugelis it only stuttering during buffering or the whole time?
17:56:08TheSevenit never manages to buffer more than ~40KB
17:57:21TheSevenflac manages to read ~300KB ahead once and then also starts rebuffering like every second
17:58:19*TheSeven wonders if there are some tunables he could play with
18:00
18:02:16 Join Buschel [0] (~chatzilla@p54A3B3F6.dip.t-dialin.net)
18:02:27TheSevenhm, the disk can barely keep up with flac apparently
18:03:02TheSevenif i remove all yields from the ata driver, it manages to read up to ~2MB ahead
18:03:13n1swhat was it you needed someone to test to be able to commit your work?
18:03:26TheSevenin complex areas of the song the usefl value drops badly
18:03:46TheSevenit has barely ~1MB of usable data after a total of ~20MB played
18:04:10TheSevenn1s: disk access on all big-endian or ata-swapped-words targets
18:04:39n1si can test on my h300, i'ts big endian, i have no idea if it swaps words though
18:04:39TheSevenfirst rebuffering at ~25 of ~27MB played this time
18:05:24TheSevenand MP3 is worse than before, even though it manages to buffer a bit more data now (this time the codec seems to be starving, not the disk)
18:05:57TheSevenno rebuffering so far, ~500KB readahead accumulated during the first ~3MB
18:06:00n1swhere can i find the patch?
18:08:05TheSevenn1s: http://pastie.org/1423647
18:08:28n1sanything special or just any reading/writing?
18:08:36TheSevenif it fails, nothing will work at all
18:08:57n1sok
18:09:10TheSevenhm, mp3 even fails after it has finished buffering (at ~90% of the track)
18:12:42TheSevenhow does the ATA driver tell the low level driver which PIO mode it should use?
18:16:19n1sTheSeven: something is definitely broken, menu and filebrowser text is scrambled
18:16:30TheSevendamn
18:16:58TheSevenbut it managed to mount the volume?
18:17:11n1slooks like it
18:17:19TheSeventhat's interesting
18:18:07TheSevenTorne: do i understand correctly that rockbox always uses PIO0?
18:24:08TheSevenapparently now
18:24:09***Saving seen data "./dancer.seen"
18:24:11TheSevennot*
18:30:00n1sTheSeven: think i found one bug in your patch
18:31:19n1sIIUC it handles all coldfires the same even though some do the swap words thing
18:31:50 Join Kitar|st [0] (~Kitarist@BSN-182-77-193.dial-up.dsl.siol.net)
18:32:42TheSevenso one would need more #ifdef'ing in firmware/target/coldfire/ata-target.h?
18:32:54n1syes, i think so
18:33:00TheSevenMPIO is handled by a different file
18:33:20n1soh, i missed that then
18:37:25 Quit feisar- (Ping timeout: 240 seconds)
18:41:47 Quit kevku (Quit: KVIrc 4.0.2 Insomnia http://www.kvirc.net/)
18:42:06n1sTheSeven: removing the swap16 from ata.c:1141 fixes it
18:42:26 Join kevku [0] (~kevku@arch.tunnel.ipv6.estpak.ee)
18:43:53TheSevenhm, but that seems to be needed on other targets
18:44:28TheSevenwhich raises the question which targets need it and which don't
18:46:15n1sthat seems odd since before the patch only big endian targets that don't swap words had this swap (and now that ATA_IN16 already swaps for those targets that extra swap was just swapping back)
18:48:14n1sIIUC an extra swap there would cause wildly different settings to be used by the ata driver
18:51:35 Quit Llorean (Quit: Leaving.)
18:55:24TheSevenit works fine on my ipod classic - but maybe that's specific to that platform
18:56:57*pixelma curses mpegplayer debug output in a sim for a file which only plays without audio in Rockbox but works fine in VLC
18:57:21pixelmathere's so much info that I can't even see what actually happens :\
18:57:39Buschelcan you see which audio codec is used?
18:58:53pixelmamaybe that's a hint: audio: mad_header_decode failed:forbidden bitrate value ?
18:59:42Buschelwhat does vlc tell you?
18:59:51n1shmm, a dump of the ata identify info with svn is different from one with the patch and my change
19:00
19:01:02TheSevenhow different?
19:01:29pixelmathe video is a music video where I had to transcode the video part but it has mpg audio, so I thought I could do without transcoding - it's 22.05 kHz though and 48 kb/s. Is mpegplayer not able to resample?
19:02:23n1sTheSeven: i just md5sumed them, checking it out now
19:04:50 Join y4n [0] (y4n@unaffiliated/y4ndexx)
19:05:10Buschelpixelma: I don't know... but good question
19:05:35n1sTheSeven: 2 bytes differ
19:05:40Buschelpixelma: which layer is it?
19:06:17amiconnn1s: Nothing interesting in there. config.log has been touched the last time at 14:32 (build started at 14:30), the build error happened 14:55
19:06:20y4nHello. It is weird how with a certain loud album I can hear volume fluctuating during listening, on a rockbox'd clip+. Compressor is off... no such issues when listening in foobar2000 for example.
19:06:35y4nenvy - Recitation (2010)
19:07:16n1samiconn: ah, ok, i set up cygwin on my netbook to try and figure it out
19:07:58amiconnUmm, on a netbook it will probably take *very* long (assuming netbook == atom)
19:08:33n1syes
19:08:34pixelmaBuschel: VLC stats tell me that there are also "lost buffers" (? losely translated back from its German UI) right at the start. I don't know what that means though
19:08:47pixelmaI'll try something else
19:09:17n1sTheSeven: in fact 4 bytes differ
19:09:17pixelmain the audio part
19:09:54n1sbut that means the swap is wrong for h300 at least
19:09:56linuxstbpixelma: Can you upload this somewhere?
19:10:40TheSevenis it normal that the pcm buffer never gets filled more than ~60%?
19:11:12 Join feisar- [0] (jljhook@irkki.fi)
19:11:28amiconnTheSeven: There's an easy way to find out whether buffering is causing the slowdown: Start playback, then press pause immediately, and wait until the disk stops. Only unpause after that
19:11:36Buschelpixelma: what happens if you simply comment line 167 in libmad/frame.c ?
19:12:24pixelmalinuxstb: I first wanted to try if my transcoding program does something weird - I also cut the file because it has parts I'm not intersted in at the beginning and the end. I'll try without the cutting
19:12:45TheSeveneven when it's finished buffering CPU usage is way too high
19:12:54TheSevenit's fine in test_codec though
19:13:52TheSevenI have ATA DMA up and running now - it manages to read about twice as fast as it plays for FLAC now
19:14:27pixelmaBuschel: layer 3
19:15:19TheSevenMP3 still isn't realtime
19:15:53BuschelTheSeven: what is the DRAM / SRAM speed like?
19:16:15 Quit feisar- (Ping timeout: 276 seconds)
19:16:23TheSevenno idea
19:16:34 Join kadoban [0] (~kadoban@ip98-165-177-158.ph.ph.cox.net)
19:16:42TheSevenbut the test_codec results seem to be sane
19:16:44Buschelcan you check with test_mem?
19:16:49TheSevenhttp://pastie.org/1423556
19:17:19TheSevenall >100MB/s
19:17:23Buschellooks good. are the decoding times correct?
19:17:27 Join balintx [0] (~quassel@szerver1.gulyasp-koll.sulinet.hu)
19:17:41TheSeventhey should be, at least roughly
19:17:43Buschel"all >100 MB/s" is too high imho
19:17:50n1samiconn: my build failed with some libiconv error
19:17:58 Join WonTu [0] (~WonTu@p57B572F3.dip.t-dialin.net)
19:18:04TheSeventhe tick might be off by a few percent, but not terribly much
19:18:12 Part WonTu
19:18:30Buschelif so, the memory bandwidth would be surprisingly high
19:18:38amiconnn1s: Btw, the binutils build errored while building gprof
19:19:22amiconngprof has a separate config.log, but I can't find anything interesting in there either
19:19:27TheSevenDRAM cnt: 2048 size: 64MB, read: 104.9MB/s (610ms), write: 220.6MB/s (290ms), memset: 220.6MB/s (290ms), memcpy: 133.3MB/s (480ms)
19:20:17TheSevenIRAM cnt: 2048 size: 64MB, read: 128.0MB/s (500ms), write: 182.8MB/s (350ms), memset: 182.8MB/s (350ms), memcpy: 152.3MB/s (420ms)
19:20:59TheSeven108 iterations so far
19:21:17pixelmalinuxstb: a file produced without the cuts works
19:22:10TheSeven~2.2 seconds per iteration, so the msec values seem to be sane
19:22:15Buschelthese numbers are quite high. I like to see the IRAM speed reaching the speed of the PP5022.
19:22:47*TheSeven wonders how to leave this plugin
19:22:56Buschelskip left
19:23:08TheSevendoesn't do anything
19:23:18Buschelwait a while
19:23:38Buschelotherwise press Menu + Select for while ;)
19:23:40TheSevenno reaction even after several iterations
19:23:44 Join feisar- [0] (jljhook@irkki.fi)
19:23:45TheSevenhm, that's what i'm about to do
19:24:46*amiconn retries the build, in order to exclude random failure
19:25:11n1samiconn: i'm rerunning my build too
19:25:52TheSevenany other ideas what might be causing playback to eat cpu like shit?
19:26:08Buscheldsp stuff
19:26:21*TheSeven doesn't think there is dsp stuff enabled
19:26:26Buscheldid you try to run test_codec including dsp
19:26:28Buschel?
19:26:31TheSeventhe PCM driver should only cause one IRQ per ~9000 samples, if the chunks passed by the pcm buffer are that big
19:26:42n1sTheSeven: does the port define CPU_ARM and ARM_ARCH?
19:26:48Buschelit is copying all pcm data from a to b
19:27:22TheSevenn1s: i'd think so, as the test_codec results seem to be reasonable
19:27:50n1sright
19:28:11TheSeventest_codec with DSP is also several hundred percents realtime, even for MP3
19:28:17Buschelok
19:28:19 Quit feisar- (Ping timeout: 264 seconds)
19:28:26TheSeven430.70% to be exact
19:28:33TheSeven44.5MHz
19:29:18TheSeven(320kbit/s)
19:29:32 Join feisar- [0] (jljhook@irkki.fi)
19:29:43TheSevenflac: 1032.30%, 18.56MHz
19:30:16*TheSeven tries test_disk again
19:30:57TheSevenstill only 16KB/s for 512 byte writes
19:31:12TheSeven512 byte reads are at ~1.5MB/s
19:32:09TheSevenwrites are about twice as fast as reads for bigger chunk sizes, weird... hm, write caching...
19:32:30TheSevenanyway, this shouldn't be the bottleneck any more
19:32:47TheSevenreads are ~1.6MB/s independent of the size
19:33:17TheSeventhat's way slower than i'd expect, but should still be sufficient for playback
19:33:33 Quit feisar- (Ping timeout: 240 seconds)
19:33:41TheSevenso this could be:
19:33:55TheSeven- some kind of fight between threads
19:34:02TheSeven- some buffering/playback issue
19:34:05TheSeven- the pcm driver
19:34:17TheSevenany other ideas what could be the culprit that doesn't influence test_codec?
19:34:58linuxstbpixelma: My guess is that the mp3 frames are cut, meaning the audio stream doesn't start on a frame boundary. That may or may not be legal, but I would expect vlc to be a lot more tolerant of such things than mpegplayer.
19:35:04TheSevenis PCM-WAV significantly faster than FLAC?
19:35:09TheSevenif yes I might try that
19:35:32n1sTheSeven: it should be
19:36:03n1sstraight 16 bit 44.1kHz pcm needs basically no processing at all
19:36:25TheSeven*if* it does no processing at all
19:36:36pixelmalinuxstb: sounds likely. I wonder if mpegplayer should also tolerate this or not. Of course you can't take care of every corner case, I just pity that it looks like I'll have to transcode the audio
19:37:07pixelmamaybe there are better transcoders though which can handle this correctly
19:37:46pixelmaand cut on frame borders
19:38:14TheSeven2.38MHz, 8039.57% realtime
19:40:03TheSevenand guess what? it's stuttering!
19:40:39TheSevenso the two "extremes" wav and mp3 are stuttering, while flac works. what the hell?
19:41:07TheSeventhe UI is barely responsive while playing the WAV file
19:41:19 Join feisar- [0] (jljhook@irkki.fi)
19:42:59linuxstbpixelma: Do you have access to mplayer? You can use that to dump the audio stream (mplayer -dumpaudio -dumpfile audio.mp3 file.mpg)
19:45:00TheSevenwav even stutters after buffering completed
19:45:07TheSevenapparently every time the wps updates
19:45:42Buschelthe screen update?
19:45:48TheSevenmaybe
19:46:01TheSevenand if buffering hasn't completed, it just reads one little chunk from the disk once a second, judging from the disk's noise
19:46:05Buschelhow much fps do you get via test_fps?
19:46:06n1sTheSeven: does the lcd update disable interrupts for too long or something
19:46:13TheSevenit shouldn't
19:46:57Buschelwhat happens if you play a file while being in the main menu?
19:47:10n1samiconn: my binutils build now completed... the error i got before went away when i installed libiconv but it looked different from your error
19:47:29TheSeventest_fps data aborts at yuv
19:47:42TheSevenrgb is 25 full / 100 quarter
19:49:02amiconn25fps is awfully slow
19:49:33Buschelafaik wps did 5 fps (full screen). so, this would equal 20% of the cpu time during playback. if the 5-fps-full-screen assumption is still valid
19:49:36TheSeventhis is still non-optimized C code, and the LCD is rather big (320x240x16bits)
19:50:26 Quit kugel (Remote host closed the connection)
19:50:56TheSevenjudging from the stuttering and lcd contents it's just doing a single frame per second (cabbiev2)
19:51:25TheSevenand the stuttering doesn't change if i go to the main menu
19:52:14Buschelthen the screen update are not the problem. main menu is not drawn cyclic
19:54:53n1shmm, the gmp/mpfr kluge in rockboxdev.sh needs to be updated for 4.5 as it needs mpc too
19:54:56TheSevenand if flac plays fine, i can't see any reason why wav shoudln't, if it's not buffering
19:55:23amiconnn1s: binutils build worked here too. weird.
19:56:18*TheSeven is running out of ideas
19:58:10 Join GeekShad0w [0] (~Antoine@93.21.168.55)
20:00
20:00:15Buschel[Saint]: you there?
20:00:39 Quit GeekShadow (Ping timeout: 272 seconds)
20:02:28 Join GeekSh4dow [0] (~Antoine@93.21.167.104)
20:03:49 Quit GeekShad0w (Ping timeout: 240 seconds)
20:06:00TheSevenhm, our bigendian targets are all the archos, iriver, iaudio, mpio and meizu players, and additionally the yp-s3?
20:06:05TheSevenwhich of those have ata?
20:06:20TheSevenata_swap_words targets seem to be only archos and mpio
20:07:36TheSevenn1s: does that mean that that your iriver is one of those identify-swap-needed targets (bigendian but not ata_swap_words)
20:09:36n1si think so, yes
20:10:09TheSeventhat's a bit confusing...
20:11:56TheSevenso on iriver, the read is already swapping, and that's sufficient
20:13:08TheSevenon mpio the read normally isn't swapped, but ata_swap_words is defined
20:13:23TheSevenso the identify info won't be swapped at all
20:13:35TheSevenso removing the additional swap would handle that as well
20:14:16TheSeventhe same holds true for archos
20:14:35TheSevenmeizu doesn't have ata-based players afaik, and the yp-s3 is flash-based as well
20:15:04TheSeveniaudio seems to work like iriver
20:16:20TheSevennone of the littleendian targets have ata_swap_words, so they never swap anything
20:16:34TheSevensounds like that swap16 should be removed, as the ipod classic is the only one needing it
20:17:00TheSevenso the ipod classic is actually a littleendian ata_swap_words target?
20:17:11 Quit jfc^2 (Read error: No route to host)
20:17:35 Join jfc^2 [0] (~john@dpc6682208002.direcpc.com)
20:18:50TheSevenata_swap_words doesn't seem to affect data transfers on bigendian platforms
20:20:00TheSevenso the data endian swap actually happens for all bigendian targets
20:20:32TheSevenwhich means my patch is broken on mpio and archos
20:21:51amiconnlittle endian targets should never need ata swap because ata is also little endian
20:22:12amiconnBig endian targets do need it, unless there's some hardware swap mechanism
20:22:42TheSevenit seems like there is a little endian target with a hardware swap mechanism :/
20:23:09amiconnThat's weird
20:23:15TheSeventhe identify words i'm getting on the ipod classic are big endian
20:24:05pixelmalinuxstb: I made something which I think is the audio dump with the transcoder I use but I'm not sure if it really leaves the stream "in peace"
20:24:11***Saving seen data "./dancer.seen"
20:26:58 Quit GuySoft (Ping timeout: 248 seconds)
20:27:16 Quit bmbl (Quit: Verlassend)
20:29:31TheSevenamiconn: any idea why the data endianness is swapped on all bigendian targets, but the identify endianness only on iriver and iaudio?
20:29:36TheSevenor am i misunderstanding something?
20:30:07TheSevenare some OFs writing data with wrong endianness, and we want to be compatible with that?
20:30:41TheSevennot that it would make any sense...
20:32:57 Quit stoffel (Ping timeout: 276 seconds)
20:38:35linuxstbpixelma: I don't know if it's your transcoder, but that audio stream has an ID3 tag at the start. That would confuse mpegplayer...
20:39:03pixelmaI wondered why there even is one in the file
20:39:29pixelmafound it too when playing the dumped mp3 in foobar
20:40:44pixelmaare you sure it's a tag? I wonder why an uncut file would play if the transcoder would add it always
20:40:51 Quit Buschel (Ping timeout: 240 seconds)
20:42:51linuxstbpixelma: Yes, looking at it in a hex editor shows it begins with the string "ID3"
20:44:00TheSevenamiconn: if your statement above is correct, this comment would be very confusing:
20:44:01TheSeven this info differently that normal sector data */
20:44:09TheSeven /* the IDENTIFY words are already swapped, so we need to treat
20:44:09TheSeven this info differently that normal sector data */
20:44:23 Quit GeekSh4dow (Quit: The cake is a lie !)
20:44:43n1shmm, the mpc lib isn't available from the same mirrors we get gmp and mpfr
20:44:51pixelmaTheSeven: its grammar is confusing anyway ;)
20:45:29 Join Judas_PhD [0] (~kevin@misterfluffy.dsl.xmission.com)
20:46:55TheSevenoh, this is getting even more confusing:
20:47:09TheSevenwhat about the strings in the identify info? which byte order do those have?
20:47:45TheSevenif i swap the byte order, the strings look right in the hexedit
20:48:09TheSevenwhat does this mean for the endianness of the rest?
20:54:57TheSevenalso, if I undestand page 114 of ATA8-ACS right, ata isn't hardwired to 512 bytes per sector
20:56:50TheSevenor rather pages 23, 113 and 114
20:57:26GodEaterhahaha
20:57:30GodEaterthis sounds like where I came into Rockbox
21:00
21:01:07 Join Buschel [0] (~chatzilla@p54A3B3F6.dip.t-dialin.net)
21:09:25 Part y4n
21:10:20 Quit balintx (Remote host closed the connection)
21:13:24 Quit factor (Quit: Leaving)
21:14:18 Join factor [0] (~factor@75.108.68.114)
21:14:22 Join balintx [0] (~quassel@szerver1.gulyasp-koll.sulinet.hu)
21:17:23 Quit factor (Read error: Connection reset by peer)
21:17:38 Join factor [0] (~factor@75.108.68.114)
21:22:10ThopterI'm new to Rockbox, having installed it via the Rockbox Utility on my Clip+ 8GB earlier today. I found a language quiz patch that I would like to install, but I'm not sure how to go about doing that. All the info I've read about adding patches talks about using diff files, but this patch is provided as a .c file. Can someone point me to some instructions somewhere that could get me through?
21:28:54BuschelThopter: can you compile rockbox?
21:29:31ThopterBuschel: with a decent set of instructions to get me started, sure
21:30:14BuschelThopter: then you should first walk through this http://www.rockbox.org/wiki/HowToCompile
21:32:54ThopterBuschel: Ok, that looks straightforward enough. What do I do with this .c patch file?
21:33:24Buschelis this a patch or a new plugin?
21:33:32Thopterit's a plugin
21:34:35 Join simonrvn_ [0] (simon@197.159-ppp.3menatwork.com)
21:34:37Buschelis it in the tracker somewhere (I am talking of flyspray)?
21:35:07Thopterflyspray 2896
21:35:47[Saint]anyone tried building convttf lately? I have a binary already compiled...so I can still use aliased fonts, but trying to build convttf fails quite dramatically.
21:36:48BuschelThopter: it's a diff file. this patch does not apply anymore?
21:37:08ThopterI downloaded the .zip file, inside was a .c
21:37:21BuschelThopter: check the diff file below
21:37:32 Quit simonrvn (Ping timeout: 240 seconds)
21:37:32 Nick simonrvn_ is now known as simonrvn (simon@197.159-ppp.3menatwork.com)
21:38:16[Saint]Buschel: Sorry about last night, I had a really bad headache and went to bed. Building with latest LCD patch now.
21:38:46Buschel[Saint]: no prob, I hope your feeling ok now :)
21:38:56Buschelyou're
21:39:06ThopterBuschel: ah, I see. The fellow's instructions aren't very clear to me though.
21:40:14BuschelThopter: what is not clear?
21:41:08Thoptergiven that reaction, something that should be clear. guess I'll go read some more
21:42:40BuschelThopter: you should set up and development environment and downlad the source. then compile without any patch and see whether this is working. if so, you may apply the patch (as described in flyspray) and rebuild.
21:43:25Thopter... yeah, I need to read up some more on this. Thanks for the assistance
21:43:27OghamCan anyone throw any light on bug FS #11812 ? Does this have the potential to damage low impedance IEM's?
21:43:48BuschelThopter: you're welcome
21:44:44TheSevenOgham: i doubt that it will destroy them thermally
21:44:55Thopterany chance this plugin (2896) might make it into the program someday?
21:45:14TheSevenloud noise should be worse for them
21:45:21*TheSeven killed a pair that way recently
21:46:20[Saint]it could, potentially tear the speaker cone if the sample it paused on was incredibly loud, and the volume was very high.
21:46:51BuschelThopter: I don't think so. It is in flyspray since ages without being included to official builds.
21:46:52*TheSeven doubts it will tear it more than playing back that sample without pausing
21:47:33TheSevenbtw. i'm sure other targets have the same problem
21:47:34Thopterso I'll have to do this for every update. Oh well, first time is always the hardest
21:47:57[Saint]Thopter: I'd be quite surprised if it applies cleanly.
21:48:01[Saint]Or, at all.
21:48:13Thopterbecause it's old?
21:48:14[Saint]it is ~4 years old.
21:48:25OghamOwch, apparently the clip+ has been measured putting out 240mV whilst paused, do you think that would be enough to damage 16 Ohm Impedance IEM's?
21:48:56Buschel[Saint], Thopter: one step after the other. imho the patch itself is quite simple. so, it should be easily adaptable to the current code
21:49:33[Saint]Oh yes, it's certainly able to be applied.
21:49:37 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
21:49:55[Saint]It just depends on the level of effort one is prepared to go to to get it to do so.
21:50:03[Saint]And the knowledge of the person doing so.
21:50:11TheSevenOgham: that would be ~4 milliwatts of power
21:51:48n1sThopter, [Saint]: a plugin that old will not build due to the plugin header changes
21:52:17n1seasily fixed though, just delete the old header macro in the plugin .c file
21:53:09OghamTheSeven: I'm not too hot on my basic physics! But that doesn't seem like much power.. do you think it would cause issues if it was constant for a long duration?
21:53:45[Saint]check the specs of your IEMs
21:53:52TheSeveni think the only impact a long duration can have is heat. and 4 milliwatts of heat shouldn't be an issue.
21:54:51TheSevenit might be possible that that voltage is sufficient to bend the membrane excessively, but it won't make a difference if that's applied for a second or an hour in that case
21:58:29 Quit feisar- (Ping timeout: 240 seconds)
21:59:36OghamTheSeven: Ah, ok.. and if 240mV applied during playback does not cause an issue, then I am I correct in assuming that it shouldn't cause an issue if applied constantly whilst the device is paused?
21:59:57TheSeveni didn't say that
22:00
22:00:27soapBuschel, what do you want tested? [Saint] you testing too?
22:00:28TheSevenif alternating voltages follow each other in rapid succession, the membrane won't be bent as far as when a constant voltage is applied
22:00:54[Saint]soap: Yessir, building now.
22:01:13TheSevenso 1/50th second vs. 1 second makes a difference, while 1 second vs. 1 hour probably won't
22:01:29TheSevenpersonally i'd expect them to survive that though
22:01:42OghamAh, that makes sense..
22:02:18OghamIf they were to be damaged in the manner what is the likely outcome? An instantly recognisable difference in sound quality, or a slow degredation?
22:02:21Buschel[Saint], soap: if the latest patch works for you both I will submit. afterwards I have some further changes to verify ;)
22:03:20*[Saint] isn't sure if this discussion is still on topic or not.
22:05:06soapdefine "latest". That's really the question I was getting at. Are you going to commit the last one I tested, and want to verify the changes in the most recent FS patch, or is the most recent FS patch the one you want to commit and thus needs my testing?
22:05:33gevaertsTheSeven: I'm assuming you use SECTOR_SIZE=512?
22:05:40TheSevenyes
22:05:45gevaertsAnd you have MAX_LOG_SECTOR_SIZE defined to something
22:06:24Buschelsoap: v17 (=latest) needs to be tested by Saint. as this (hopefully) fixes some issues he had on his iPod color
22:06:24TheSeven#define MAX_LOG_SECTOR_SIZE 4096
22:06:24TheSeven#define MAX_PHYS_SECTOR_SIZE 4096
22:06:35gevaertsok
22:07:25Buschelsoap: afterwards I will make further changes I would like to have tested by you both
22:07:52gevaertsThe problem is that usb_storage use disk_sector_multiplier, which is detected by disk.c based on which "actual" sector sizes it manages to mount partitions with
22:08:15gevaertsFor superfloppy mode it never sets it
22:08:30TheSevenaha, sounds like that should be fixed :)
22:08:49soapso, not to beat a dead horse Buschel, to be perfectly clear v17 does not need testing by soap.
22:08:57gevaertsPresumably disk_mount() could set disk_sector_multiplier for superfloppy mode based on the fat bpb_bytspersec field
22:09:11OghamYup sorry to be off-topic! I'll shut-up now, I'm just really worried about damaging my hardware due to this bug!
22:09:11 Quit Thopter (Quit: Sayonara)
22:09:48Buschelsoap: yes, v17 does not to be tested by you.
22:10:17OghamSo last question.. if this bug was going to cause damage to my IEM's would it be instantly recognisable or a slow degredation?
22:10:30*Ogham promises not to ask any more questions! :)
22:10:41[Saint]Well, at the end of the day the no warranty/guarantee is pretty specific.
22:10:54[Saint]And, that question is almost impossible to answer.
22:10:56*TheSeven thinks nobody knows a definitive answer to that, besides the manufacturer maybe
22:11:26soapOgham, IF IF it were to cause damage it would likely be through voicecoil overheating and eventual fuzing.
22:11:40soapIF IF IF IF
22:11:44TheSevengevaerts: so the fat code relies on bpb_bytspersec anyway?
22:12:03Buschel[Saint]: still compiling?
22:12:06gevaertsI guess so
22:12:19TheSevensoap: voicecoil overheating won't happen at that wattage, i would be more concerned about mechanical membrane damage
22:12:37[Saint]Yes, sorry. I needed to clean my build dir so it's taking longer than normal.
22:12:42[Saint]Buschel: ^
22:12:55soapthat voltage is too low as a percentage of "expected" to see much distention, no?
22:13:00 Quit factor (Ping timeout: 246 seconds)
22:13:55soapI mean, if a 10v rail to rail headphone amp won't bend the membrane too much why would 1/4v?
22:16:15TheSevena 10v rail to rail headphone amp? i doubt one will use such a thing at 16 ohms impedance
22:16:48 Quit saratoga (Quit: Page closed)
22:17:05TheSeveni'm pretty sure that one *can* damage 16ohms inears at 1V
22:17:59TheSeven10V at 16 ohms would be 6 watts. that will certainly kill the voice coil.
22:18:24 Join AlexP_mob [0] (~ap@rockbox/staff/AlexP)
22:19:54gevaertsTheSeven: http://pastebin.com/SKp6KYVb (compiles, but untested apart from that)
22:20:25*TheSeven will throw that into his tree :)
22:23:44gevaertsTheSeven: actually, it might be interesting to see what the OF does with a plain standard 512 byte sector drive
22:24:08gevaertsIf it also treats that as 4K, it might be better to use a fixed value
22:24:12***Saving seen data "./dancer.seen"
22:24:20OghamThanks very much for your time all, one last thing then.. taking the no guarantee/warranty as a given - would you all be happy to continue usingexpensive IEM''s with rockbox in this situation?
22:24:54[Saint]I use a pair of Sennheiser IE8s
22:25:00*gevaerts considers using expensive IEMs to be proof of insanity, so he wouldn't use them, including in this situation
22:25:09[Saint];)
22:25:34OghamTrue.. I wish I had bought cheaper ones I would not be so concerned about now!
22:25:35TheSevengevaerts: i seriously doubt we can set sector_size to 4k without major ata rework
22:25:53gevaertsTheSeven: not sector_size, no, but hardcode disk_sector_multiplier
22:26:04Ogham[Saint]: Do you use them on a Clip+ with this firmware bug?
22:26:19TheSevenhardcoding that won't buy us much
22:26:38TheSeven(except from rejecting partitioning that the OF wouldn't like)
22:26:44gevaertsThe way things work now is that we actually have to trust data on the disk (partition table, FAT bootsector,...) to tell us how to interpret that same data
22:27:07[Saint]No, but at the end of the day I think you're being a little bit over cautious. A meteorite could plummet to earth and smash your IEMs, things break.
22:27:23AlexP_mobThat is a bit silly
22:27:47[Saint]That was kind of the point.
22:27:48AlexP_mobYou can't avoid that, you can this
22:27:55gevaertsOgham: have you considered contacting the manufacturer of your IEMs? They're much more likely to know
22:28:07gevaertsWe can really only guess
22:28:28AlexP_mobSaint no, I meant your point was silly
22:28:49[Saint]I realise that.
22:28:53gevaertsIf the things are really expensive, I'd assume they do have a support service that can be reached :)
22:29:33Oghamgevaerts: I think I better had.. that or switch back to OF and hope that I have not already caused considerable damage!
22:29:40TheSevenOgham: you could just put a decoupling capacitor in between)
22:29:45gevaertsAre you sure the OF doesn't do this?
22:30:01Oghamgevaerts: definately, its a rockbox issue
22:30:16gevaertsSure, but that's not what I asked
22:30:25OghamTheSeven: I guess I could try that.. again I am a little restricted by my lack of technical knowhow!
22:30:48gevaertsThe fact that rockbox has a bug does not prove anything about the OF
22:31:05[Saint]The forum suggests it's rather hit and miss, and dependant on the IEMs.
22:31:16[Saint]As I understand it, not all are experiencing this.
22:31:31Oghamgevaerts: Well, the mV was measured as 0 when the player was paused and using OF
22:31:37gevaertsok
22:31:45gevaertsTested multiple times, I presume
22:32:24gevaertsI mean, with this sort of bug, the value you get is more or less unpredictable and will be zero every now and then
22:32:30Oghamgevaerts: Yes, people suggested possible problems with the tests such as no load, and the tests were repeated
22:33:41[Saint]Buschel: success...stand by for FPS results.
22:33:50Buschel\o/
22:35:00Ogham[Saint]: well, the voltage is always there on a meter, but different IEM's vary in if they hear a squeak when the player is paused/resumed
22:37:10[Saint]LCD 1/1: 19.4, 1/4: 77.1 LCD YUV 1/1: 17.4, 1/4: 70.0 @30MHz ; LCD 1/1: 51.7, 1/4: 206.0 LCD YUV 1/1: 46.7, 1/4: 187.0 @80MHz
22:37:14[Saint]Buschel: ^
22:38:54amiconnTheSeven: A, that's a detail I forgot. For some reason identify data is big endian, unlike normal ata data
22:39:00OghamWell, I think I better switch back to the OF for the time being.. I'm not convinced that my IEM's are safe, I just hope they are not already damaged too much!
22:39:11OghamThanks again all, sorry to get off-topic here!
22:39:25TheSevenamiconn: ah? i had the opposite impression when looking at the code
22:40:16amiconnSo if you want the 16 bit ints to be little endian, you need to byte-swap - but then strings are in the wrong order
22:40:50amiconnActually, that's probably the reason why it's big endian
22:41:21*TheSeven is completely confused now
22:42:18[Saint]Buschel: Where is the patch for the LCD shotdown artefacts I'm getting on iPod Colour? It's not on the tracker.
22:43:05TheSevenamiconn: apparently, on little-endian targets, identify and data is handled the same way, while on big endian targets they're handled differently
22:43:19CIA-7New commit by 03Buschel (r28944): Submit FS #11843 v17. Integrate YUV-blitting of nano 2G to nano1G/color LCD driver. Additionally refactor RGB and YUV screen updates to use same code ...
22:43:38[Saint]oh...found it.
22:43:52[Saint]Oh, wait...nope :/
22:43:55 Join GuySoft [0] (guy@bzq-79-179-40-185.red.bezeqint.net)
22:44:25[Saint]Buschel: Do you have the wait you added to LCD in a patch form?
22:45:57[Saint]aha, found it in the logs...sorry Buschel.
22:46:02[Saint]I'll test that now.
22:47:50 Quit Judas_PhD (Ping timeout: 240 seconds)
22:55:20 Join feisar- [0] (jljhook@irkki.fi)
22:58:42Buschelsoap, [Saint]: can you now check this patch? -> http://pastie.org/1424222
22:58:48Buschel(I call it v18)
22:59:28[Saint]what does this one do?
23:00
23:00:10Buschelit removes some (from my understanding unneeded) special handling
23:00:14Buschelin the LCD driver
23:00:30 Join Judas_PhD [0] (~kevin@misterfluffy.dsl.xmission.com)
23:01:01soappatch against current svn?
23:01:12soapduh, yes
23:02:48 Quit feisar- (Ping timeout: 265 seconds)
23:03:02 Join feisar- [0] (jljhook@irkki.fi)
23:03:42 Join Keripo1 [0] (~Keripo@CPE0022b0d4bdb7-CM001a6680d4fe.cpe.net.cable.rogers.com)
23:04:20 Quit Keripo (Ping timeout: 264 seconds)
23:09:25[Saint]Buschel: The screen artefact patch doesn't work.
23:09:47[Saint]It just displays a full white screen before the horrible vertical lines now :/
23:09:48Buschelok, was a quick shot..
23:09:50[Saint]Sorry.
23:10:14[Saint]It disturbs me how long the artefacts last for.
23:10:20 Quit bertrik (Ping timeout: 255 seconds)
23:10:36[Saint]I would swear that it wasn't being powered down properly, as sometimes they can last on screen for over 5 minutes.
23:13:54*[Saint] has nice boot-splash and USB icons for his iPod Colours now.
23:15:44*TheSeven thinks he's caught all the corner cases of that ATA patch now
23:16:05TheSevendoes anyone want to (re-)test it before i commit it?
23:16:32n1ssure
23:19:51TheSevenn1s: http://pastie.org/1424257
23:20:47soapBuschel, no speed changes no regression in quality on Nano 1G.
23:21:18Buschelsoap: good. let's hope Saint will report the same.
23:22:12[Saint]It will be a while before I can test this latest increment, but...fingers crossed, yes.
23:22:16Buschelsoap: btw, could you update the LcdFrameRate wiki or give me the detailed numbers for all measurements (YUV/RGB/boosted/unboosted)?
23:22:26soapon it
23:22:40Buschel[Saint]: shall I upload a binary for you?
23:22:51[Saint]Buschel: I will do the same for the Colour when I've finished the test.
23:23:29[Saint]Buschel: It's building now, it's just a timing thing. I have to run out the door for about an hour shortly.
23:23:40Buschelok
23:24:26 Quit TheSeven ()
23:24:32 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
23:24:37amiconnTheSeven: This isn't easy to understand. Afaiu the different handling of identify and plain data on big endia is because we want plain data in original order, but identify data in target endianess
23:26:05amiconnSo on little endian no swap is required for both, while on big endian targets with hardware swap support we need to swap identify but not data, and on big endian without hardware swap support we need to swap data but not identify
23:26:37TheSevenyeah, i think i got it
23:27:20TheSevenand the ipod classic is different again: the data register is swapped, but not the control registers
23:27:24TheSevendon't ask me why
23:27:26amiconnActually hardware swap support just means ata buffer lines 0..7 are connected to lines 8..15 of the data bus and vice versa
23:28:35 Quit Judas_PhD (Quit: This is a quitting message)
23:28:38CIA-7New commit by 03theseven (r28945): Fix an ugly-looking comment
23:28:38CIA-7r28945 build result: All green
23:28:54TheSevennow that was a fast build :)
23:29:45CIA-7New commit by 03theseven (r28946): Autodetect sector size on superfloppy volumes based on the FAT32 BPB (kudos to Frank Gevaerts)
23:29:51amiconnThat's the reason why the check_registers patterns are defined differently for the coldfire irivers
23:30:06TheSevenhm, no, on coldfire everything is swapped
23:30:09TheSevennot just the data register
23:30:11Buschelsoap: the RGB fps was not affected at all?
23:30:31amiconnMaybe the classic runs in big endian arm mode (is that possible on the SoC?)?
23:30:43soapBuschel, hasn't been affected for quite a while.
23:30:49soapI'll double check.
23:30:53soapsheesh! ;)
23:31:02TheSevennope, it's the ata core that's crazy
23:31:18TheSeveneverything else doesn't do that kind of bullshit
23:31:21Buschelsoap: Saint reported slight speedup for 1/4 screen RGB
23:31:39TheSevenapple seems to have worked around it by swapping things in software as well
23:31:57 Join Rob2222 [0] (~Miranda@p4FFF3774.dip.t-dialin.net)
23:32:01CIA-7r28946 build result: All green
23:32:25soapBuschel, yea, but all the numbers make more sense if we discount [Saint]'s findings! ;)
23:32:29TheSevenn1s: did you try the patch?
23:32:59n1sTheSeven: sorry, got stuck in a test disk run, didn't think it took this long
23:33:13TheSevenwith or without the patch?
23:33:21n1swithout
23:33:25TheSevendamn :/
23:33:35n1spaperclipped it now
23:33:46Buschelsoap: yes, true ;)
23:33:49 Quit benedikt93 (Quit: Bye ;))
23:33:54[Saint]Buschel: And what's wrong with my findings? ;)
23:34:05n1sTheSeven: no problem, it's just that the write and veryfy takes forever
23:34:57Buschel[Saint]: nothing, I better have findings before a submit than after :)
23:35:23 Join advcomp2019_ [0] (~advcomp20@97-114-234-88.sxcy.qwest.net)
23:35:23 Quit advcomp2019_ (Changing host)
23:35:23 Join advcomp2019_ [0] (~advcomp20@unaffiliated/advcomp2019)
23:35:54 Quit pamaury (Remote host closed the connection)
23:36:16 Join DerPapst1 [0] (~Alexander@91-66-226-46-dynip.superkabel.de)
23:39:01 Quit DerPapst (Ping timeout: 276 seconds)
23:39:01 Quit advcomp2019 (Ping timeout: 276 seconds)
23:39:30*TheSeven wants to get that port pushed into SVN as he won't have much time next week
23:39:40 Quit Bushmills (Ping timeout: 276 seconds)
23:41:54CIA-7New commit by 03alle (r28947): Fix typo in the comment
23:44:04CIA-7r28947 build result: All green
23:44:04CIA-7New commit by 03alle (r28948): Don't load the keyboard layout '-.kbd' (partial fix for FS #11847)
23:44:52 Quit Llorean (*.net *.split)
23:44:53 Quit kadoban (*.net *.split)
23:44:53 Quit mortalscan (*.net *.split)
23:44:53 Quit efyx (*.net *.split)
23:44:53 Quit domonoky (*.net *.split)
23:44:53 Quit chattr (*.net *.split)
23:44:53 Quit Xerion (*.net *.split)
23:44:53 Quit B4gder (*.net *.split)
23:44:53 Quit Farthen (*.net *.split)
23:44:53 Quit niekie (*.net *.split)
23:44:53 Quit tempname (*.net *.split)
23:44:54 Quit Stummi (*.net *.split)
23:44:54 Quit mc2739 (*.net *.split)
23:44:54 Quit jordan` (*.net *.split)
23:44:54 Quit linuxstb (*.net *.split)
23:44:54 Quit n17ikh (*.net *.split)
23:44:54 Quit dionoea (*.net *.split)
23:44:58 Quit AlexP_mob (*.net *.split)
23:44:58 Quit n1s (*.net *.split)
23:44:58 Quit mikroflops_ (*.net *.split)
23:44:59 Quit elcan (*.net *.split)
23:44:59 Quit simabeis (*.net *.split)
23:44:59 Quit soap (*.net *.split)
23:44:59 Quit parafin (*.net *.split)
23:44:59 Quit bug2000 (*.net *.split)
23:46:02 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
23:46:03 Join kadoban [0] (~kadoban@ip98-165-177-158.ph.ph.cox.net)
23:46:03 Join mortalscan [0] (~mortalsca@173-166-0-166-newengland.hfc.comcastbusiness.net)
23:46:03 Join efyx [0] (~efyx@lap34-1-82-225-185-146.fbx.proxad.net)
23:46:03 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
23:46:03 Join chattr [0] (~mike@244.87.189.72.cfl.res.rr.com)
23:46:03 Join Xerion [0] (~xerion@5419A4D7.cm-5-2c.dynamic.ziggo.nl)
23:46:03 Join B4gder [0] (~daniel@rockbox/developer/bagder)
23:46:03 Join Farthen [0] (~Farthen@static.225.178.40.188.clients.your-server.de)
23:46:03 Join niekie [0] (quasselcor@CAcert/Assurer/niekie)
23:46:03 Join tempname [0] (generic@server1.unitedservers.de)
23:46:03 Join Stummi [0] (Stummi@rockbox/developer/Stummi)
23:46:03 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
23:46:03 Join jordan` [0] (~jordan@jem75-13-78-235-252-137.fbx.proxad.net)
23:46:03 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
23:46:03 Join n17ikh [0] (~n17ikh@c-68-59-25-51.hsd1.sc.comcast.net)
23:46:03 Join dionoea [0] (~dionoea@videolan/developer/dionoea)
23:46:16CIA-7r28948 build result: All green
23:46:40 Join parafin [0] (parafin@paraf.in)
23:46:40 Join bug2000 [0] (~bug@unaffiliated/bug2000)
23:46:57 Join AlexP_mob [0] (~ap@rockbox/staff/AlexP)
23:46:57 Join n1s [0] (~n1s@rockbox/developer/n1s)
23:46:57 Join mikroflops_ [0] (~yogurt@h-34-71.A238.priv.bahnhof.se)
23:46:57 Join elcan [0] (user36@pr0.us)
23:46:57 Join simabeis [0] (~simabeis@lobmenschen.de)
23:46:57 Join soap [0] (~soap@rockbox/staff/soap)
23:48:44CIA-7New commit by 03alle (r28949): Don't load the colours file if it's set to '' (partial fix for FS #11847)
23:50:11CIA-7r28949 build result: All green
23:50:38 Quit kadoban (Ping timeout: 240 seconds)
23:51:07*TheSeven eagerly awaits testing results
23:51:34n1sTheSeven: seems to work fine on my h300
23:51:36 Join Bushmills [0] (~Bushmills@scarydevilmonastery.net)
23:51:42 Quit elcan (Ping timeout: 246 seconds)
23:52:01CIA-7New commit by 03theseven (r28950): Rework ATA driver to get rid of lots of target-specific constants and allow for non-memory-mapped task file registers.
23:53:14 Quit thegeek (Read error: Connection reset by peer)
23:53:29 Join thegeek [0] (~nnscript@132.108.34.95.customer.cdi.no)
23:54:00CIA-7r28950 build result: 20 errors, 8 warnings (theseven committed)
23:54:06TheSevenarrrrrrrrrrrrrrr
23:54:34 Quit scorche (Disconnected by services)
23:54:43 Join kugel [0] (~kugel@rockbox/developer/kugel)
23:54:44 Join scorche` [0] (~scorche@rockbox/administrator/scorche)
23:54:46 Quit timccc (*.net *.split)
23:54:46 Quit Sajber^ (*.net *.split)
23:54:46 Quit Q__ (*.net *.split)
23:54:46 Quit rvvs89 (*.net *.split)
23:54:46 Quit pikytcus (*.net *.split)
23:54:46 Quit scorche|1h (*.net *.split)
23:54:46 Quit fred_2 (*.net *.split)
23:54:46 Quit knittl (*.net *.split)
23:54:46 Quit Hadaka (*.net *.split)
23:54:46 Quit zu_ (*.net *.split)
23:54:46 Quit Utchy (*.net *.split)
23:54:46 Quit AlexP_mob (*.net *.split)
23:54:47 Quit n1s (*.net *.split)
23:54:47 Quit mikroflops_ (*.net *.split)
23:54:47 Quit simabeis (*.net *.split)
23:54:47 Quit soap (*.net *.split)
23:57:33 Join evilnick [0] (~evilnick@ool-18bcf602.dyn.optonline.net)
23:57:39 Quit tempname (Quit: Serverwechsel)
23:57:46 Quit evilnick (Changing host)
23:57:46 Join evilnick [0] (~evilnick@rockbox/staff/evilnick)
23:57:59 Join factor [0] (~factor@75.108.68.114)
23:58:11 Join scorche|sh [0] (~scorche@squisch.net)
23:58:14 Join Naked [0] (~naked@naked.iki.fi)
23:58:17 Join pikytcus [0] (~bigd@failbox.co.cc)
23:58:22 Nick Naked is now known as Hadaka (~naked@naked.iki.fi)
23:58:32kugelTheSeven: the ata_init() change looks suspicious
23:58:45Buschelgood night
23:58:46 Join rvvs89 [0] (rvvs89@mussel.ucc.gu.uwa.edu.au)
23:58:49 Quit Buschel (Quit: ChatZilla 0.9.86 [Firefox 3.6.13/20101203075014])

Previous day | Next day