**** BEGIN LOGGING AT Fri Apr 06 03:00:00 2012 Apr 06 08:19:47 DocScrutinizer: infobot joined! Apr 06 11:59:05 it's incredible that there are no recent slide-out keyboard smartphones these days Apr 06 12:02:08 I would like a phone running native Linux with X, and a slide out keyboard, that is pretty recent Apr 06 12:02:15 that's my only requirements for a phone Apr 06 12:02:21 is it so hard to live up to? D: Apr 06 12:02:44 N900 worked perfectly, but mine has started to really break apart recently Apr 06 12:02:58 guess it will just survive like ½ year longer Apr 06 12:03:04 max Apr 06 12:03:52 :/ Apr 06 12:29:54 hi, i want extract data from adressbook.db with perl. which dbm module do i need? (DBM_File, NDBM_File ...) Apr 06 12:36:38 sqlite Apr 06 12:36:45 it's all sqlite Apr 06 12:36:57 sqlite3 to be precise Apr 06 12:37:29 jacekowski: ok, it is possible that i have an older version in which it is DBM Apr 06 12:37:33 Can give a minimum example? Apr 06 12:37:43 jacekowski: ? no, it's libdb of some sort Apr 06 12:37:44 FIQ: I'm not that difficult Apr 06 12:38:08 FIQ: I just want a simple phone that has a good application / developer ecosystem, and a slide out keyboard. Apr 06 12:38:14 Heh Apr 06 12:38:31 And something that won't suck donkey balls as soon as you throw 3 open apps at it. Apr 06 12:38:34 i don't have n900 anymoe Apr 06 12:38:40 but i'm pretty sure it's sqlite Apr 06 12:38:44 I gave mine to my girlfriend. Apr 06 12:38:57 jacekowski: but if i try sqlite3 addressbook.db '.tables' i get the error Error: file is encrypted or is not a database Apr 06 12:39:10 .db is not a sqlite database Apr 06 12:39:21 ".sqlite" is the extension. Apr 06 12:39:25 not always Apr 06 12:39:31 look for example for calllog Apr 06 12:39:38 it's .db as well Apr 06 12:39:50 just run file on it Apr 06 12:39:58 teotwaki: addressbook.db: Berkeley DB (Hash, version 8, native byte-order) Apr 06 12:40:34 jacekowski: the conversation log is sqlite Apr 06 12:40:34 http://en.wikipedia.org/wiki/Berkeley_DB Apr 06 12:40:38 the address book is libdb Apr 06 12:40:51 but, you should access it via the eds library - maybe libebook has a perl binding? Apr 06 12:41:20 Evolution - A free/open source mail client; contacts are stored in addressbook.db using Berkley DB Apr 06 12:41:40 oh look, the same filename and the same file type! Apr 06 12:41:42 woop woop! Apr 06 12:42:44 Robot101: what is libdb? where can i find information about libdb? Apr 06 12:44:25 dsd31g: you don't really want to read the file directly - you should use libebook Apr 06 12:44:33 otherwise you will need to parse the vcards yourself Apr 06 12:45:55 Robot101: so adressbook.db is not a berkeley db? Apr 06 12:46:06 yes, libdb is berkeley db Apr 06 12:53:44 but what the abook program stores there is vcard strings Apr 06 12:53:54 so you can't just access the address book fields as a table Apr 06 12:59:08 Robot101: how vcard strings are built? Apr 06 12:59:37 some crap with lots of semicolons I think :) Apr 06 13:00:18 libebook has a vcard parser in Apr 06 13:00:43 maybe it would help if you could explain what you were trying to do? there might be some pre-existing code you could borrow from as inspiration Apr 06 13:03:23 Robot101: i want just get my data out of that fucking addressbook.db, because i haven't maemo anymore. Apr 06 13:04:01 out and into where? Apr 06 13:04:51 Robot101: in a simpler format. for example csv Apr 06 13:05:33 is your N900 dead / gone / sold / ...? Apr 06 13:05:58 my N900 is dead Apr 06 13:06:18 i actually just need the phone numbers and the names Apr 06 13:07:02 :/ Apr 06 13:08:23 i think my problem aren't the vcards. my problem is the *dbm, because i cant get the vcards out of it. Apr 06 13:12:21 its so hard to find the packages on the maemo.org website, sigh Apr 06 13:14:40 as last solution, simply open the file in an editor... Apr 06 13:18:21 Robot101: which packages? Apr 06 13:20:25 NIN102: yeah ok or with db_dump and some sed and grep. but there must be a proper way too. Apr 06 13:21:08 dsd31g: i was trying to find the maemo.org package viewer (like packages.debian.org) so I could figure out the right libdb version and maybe advise something useful there Apr 06 13:21:44 my other idea was to use eg syncevolution on the file to sync the contacts to gmail or something, but actually, there's no guarantee a desktop libebook would read the same file as maemo's libebook Apr 06 13:31:52 Robot101: thanks for the help. i will try it with db_dump now. Apr 06 13:33:09 dsd31g: dumb question - did you try on a linux desktop and just copying addressbook.db over ./.local/share/evolution/addressbook/system/addressbook.db then killing e-d-s? Apr 06 13:33:54 then some e-d-s app, eg "contacts" might see them... Apr 06 13:34:59 Robot101: ehmm... no. i don't use evolution Apr 06 13:36:33 so? just install it Apr 06 13:36:45 apt-get install contacts Apr 06 13:37:04 is a simple gtk UI for e-d-s Apr 06 13:40:30 Robot101: i'm just working on openbsd Apr 06 13:41:05 wow... er... just get an ubuntu livecd? Apr 06 13:41:35 just as a one-shot thing it might just give you all of your contacts in a GUI right away Apr 06 13:47:53 Robot101: where should i put the file? Apr 06 13:48:27 dsd31g: I just did find . -name addressbook.db Apr 06 13:48:29 in my ~ Apr 06 13:49:05 I guess killall e-addressbook-factory before putting the file in Apr 06 13:49:20 then run contacts or evolution or something, it should reload the file and... maaaaybe it will work Apr 06 13:49:24 it might also just fail entirely Apr 06 13:57:09 some OT todo: I've got to copy a partition from one harddrive to another, I though of creating a partition identical in size and then just use dd to copy the content, but I'm stuck at creating the new partition Apr 06 13:57:44 heaven't found the proper tool yet where I can input the partition size so that I'm sure it's as big Apr 06 13:58:24 merlin1991: just make it slightly larger, will work as well Apr 06 14:00:39 dm8tbr: but that feels so wrong :/ Apr 06 14:04:00 hm any ideas how big of a blocksize to use for that dd? Apr 06 14:04:03 merlin1991: if it's ext2/3/4 just run a resize2fs afterwards :) Apr 06 14:04:22 nowadays in most cases HUGE :) Apr 06 14:04:31 dm8tbr: well I managed to fit the actual logical parition to the same size Apr 06 14:04:35 * dm8tbr usually sets bs=10M Apr 06 14:04:57 only the extended one is several cylinders bigger instead of 2 like on the main disk Apr 06 14:07:05 fucking hell this is fast Apr 06 14:07:37 192 MB/s according to dd Apr 06 14:08:10 time to try to mount the new partition Apr 06 14:10:10 dm8tbr: do you happen to know how I tell grub2 to start look for it's modules over there now? Apr 06 14:10:18 s/look/looking/ Apr 06 14:10:19 merlin1991 meant: dm8tbr: do you happen to know how I tell grub2 to start looking for it's modules over there now? Apr 06 14:10:36 hum? Apr 06 14:11:24 the partition I moved contains the /boot --> the modules and config for grub Apr 06 14:11:42 now I need to tell the grub in the mbr to use this one instead Apr 06 14:11:51 dsd31g: http://talk.maemo.org/showthread.php?t=60526&page=3 Apr 06 14:13:45 DocScrutinizer: i have solved the problem. i've found out that the names and the numbers are in the el-v1.db in the Remotes table too. Apr 06 14:14:15 DocScrutinizer: so i take them from there. Apr 06 14:17:26 merlin1991: info grub ;-) Apr 06 14:17:54 iirc enters an interactive console Apr 06 14:18:48 Meanwhile: Majority of Landmark Cancer Studies Cannot Be Replicated Apr 06 14:21:08 DocScrutinizer: well it drops to a rescue shell anyway Apr 06 14:21:20 where you can locate the modules and get it running, but that isn't persistant Apr 06 14:22:18 iirc you can make it persistent in that shell Apr 06 14:28:10 though how? :D Apr 06 14:28:11 grub> setup (hd0) ?? Apr 06 14:28:28 info:/grub/Installing GRUB natively Apr 06 14:30:18 >>If you are not sure which partition actually holds this directory, use the command `find' ( find), like this: Apr 06 14:30:19 grub> find /boot/grub/stage1 Apr 06 14:30:21 This will search for the file name `/boot/grub/stage1' and show the devices which contain the file. Apr 06 14:30:22 Once you've set the root device correctly, run the command `setup' ( setup): Apr 06 14:30:24 grub> setup (hd0) Apr 06 14:30:59 no such thing as stage1 in grub Apr 06 14:31:05 only in grub-legacy :D Apr 06 14:32:34 maybe you use grub-II Apr 06 14:33:37 yeah grub2 though according to their page there is no such thing as grub2, only grub(2) and grub-legacy aka grub Apr 06 14:33:50 silly devs Apr 06 14:35:22 my manpages and ##info is for grub (one) Apr 06 14:36:02 konqueror: "##grub" Apr 06 14:36:26 or shell `info grub` Apr 06 14:37:07 (I really prefer konqueror for that particular task of reading info crap) Apr 06 14:41:19 merlin1991: >>Using the command-line interface, you type the drive specification and file name of the kernel manually. In the menu interface, you just select an OS using the arrow keys. The menu is based on a configuration file which you prepare beforehand ( Configuration). While in the menu, you can switch to the command-line mode, and vice-versa. You can even edit menu entries before using them.<< Apr 06 14:42:10 yep I know all that, but all that stuff gets lost upon actually booting an entry Apr 06 14:42:17 oops, the "edit menu entries" doesn't imply persistence Apr 06 16:46:24 http://tablets-dev.nokia.com seems to be down? Apr 06 17:35:14 Hello ! Apr 06 17:35:19 HI Apr 06 18:02:15 hello there Apr 06 18:37:14 ugh Apr 06 18:37:21 damn synology tls smtp stuff isn't working Apr 06 18:37:27 is there some other port used for smtp using tls? Apr 06 18:40:24 Macer: 465 or 587 Apr 06 18:43:44 hi there. how about compiling an existing application for use on n900? do i need the whole scratchbox stuff? madde or sdk? its just a small vpn daemon called tinc. no need for qt and all that gui stuff Apr 06 18:44:21 i would go with scratchbox Apr 06 18:46:24 i read through the 'howto' and it looks a bit complicated to me Apr 06 18:46:42 maybe i try, i really miss tinc on the device Apr 06 18:47:29 is there a working C compiler on n900? The gcc-4.6 package is empty, it would seem Apr 06 18:49:17 iirc, for that you have to activate some special repos. another way would be to use a chroot. Apr 06 18:49:49 I think I have them all, up to extras-devel. No -testing though Apr 06 18:50:11 hm Apr 06 18:50:23 there doesn't seem to be a 465 or 587 opened on the synology Apr 06 18:50:38 this must be some sort of sick joke. but then again there isn't much in way of smtp configuration either Apr 06 18:51:03 maybe the update for the dsm will be better Apr 06 18:51:16 although i always thought that tls allowed for keeping stuff on standard ports Apr 06 18:51:48 Pali: is that for ssl ? Apr 06 18:52:42 what special thing is Scratchbox Hathor ? Apr 06 18:56:08 Macer: 25 is standard SMTP port, 465 is standard SMTPS (SMPT over SSL) and 587 is port for transmission (urd?) but now used for SMTP via TLS Apr 06 18:57:53 hm i see Apr 06 18:57:55 what do people use for IMAP in n900? Apr 06 18:58:01 i'm not seeing a port 587 on the synology Apr 06 18:58:05 the standard email client does not seem to support it Apr 06 18:58:06 that sucks Apr 06 18:58:21 i should be able to connect to it with telnet and see it correct? Apr 06 18:58:24 if tls is active? Apr 06 18:58:50 i can see postfix on 25 Apr 06 18:59:13 * RST38h is impressed by how many C-grade movies have been inspired by the Da Vinci Code Apr 06 18:59:15 damn. i was hoping to avoid having to make my own email server but i guess i am going to have to since the synology based one seems limited Apr 06 18:59:28 i was messing with fengoffice heh Apr 06 18:59:33 it is pretty nice Apr 06 18:59:45 telnet does not support TLS Apr 06 18:59:55 also port 25 is used only for standard SMTP Apr 06 19:01:09 for testing SSL/TLS layer use openssl: openssl s_client -connect host:port Apr 06 19:06:53 I'm using the standard email client for imap Apr 06 19:07:12 you only see subscribed folders though and can't subscribe/unsubscribe stuff. Apr 06 19:07:18 Try it for pop3 and suffer! Apr 06 19:18:35 Pali: but at the least it would connect to the port Apr 06 19:18:38 and show garbage ;) Apr 06 19:18:47 yes Apr 06 19:18:55 i'm not getting that Apr 06 19:18:58 you can use nc for testing if you can connect to port Apr 06 19:19:14 but like i said. the config on a synology box is rather basic Apr 06 19:19:28 it is just a tick for "enable smtp" there aren't really many more options for it Apr 06 19:34:34 that you have bad email client ;) Apr 06 19:46:24 <[1]Hockmeister> is it possible to install Maemo on non-Nokia phones? Apr 06 19:47:22 No. Apr 06 19:48:34 <[1]Hockmeister> why not? Apr 06 19:50:40 Because maemo contains binary copyright elements Apr 06 19:50:52 that will ony run on the n900/n950/n9 Apr 06 19:50:57 err Apr 06 19:51:07 yeah Apr 06 19:51:35 In principle, you could make replacement for all the closed bits. Apr 06 19:51:40 But it's man-years of work Apr 06 19:52:01 and almost impossible as there's no docs about what those blobs do Apr 06 19:52:16 <[1]Hockmeister> ok. this may notbe the best place to ask this question. How would I install a different OS on a phone like LG T300? What are my options? Apr 06 19:52:26 If you're replacing _all_ the nokia copyright - then it sorta gets easier Apr 06 19:52:40 SpeedEvil: indeed Apr 06 19:52:42 :-D Apr 06 19:52:44 [1]Hockmeister: It's an enormously complex task. Apr 06 19:53:18 [1]Hockmeister: Firstly, you start out working out if it will boot your desired OS. Then you work out if drivers support all your hardware. Apr 06 19:53:22 again manyears of work Apr 06 19:53:33 Then, the software part. Apr 06 19:53:43 Indeed. Apr 06 21:10:06 nick Sicelo Apr 06 21:10:13 heh Apr 06 21:43:11 #find shazam Apr 06 21:53:43 hi Apr 06 21:53:52 trying to set profile silent Apr 06 21:53:58 but says -> /usr/bin/run-standalone.sh: line 11: /root/silent.sh: not found Apr 06 21:54:03 but its there Apr 06 21:54:13 and rights ok Apr 06 21:55:32 i wonder if that's enough info. Apr 06 21:55:43 try package 'phone-control' Apr 06 21:56:25 when I use: sh silent.sh it sets profile to silent Apr 06 21:56:40 but not with that standalone script Apr 06 21:56:54 and thats needed for using it with fcron Apr 06 21:57:48 phone-control will work fine for your needs. also, i'm not to sure i know what 'script' you're using. Apr 06 21:58:11 that standalone came with fcrontab Apr 06 21:58:27 ah Apr 06 21:58:37 * Sicelo uses alarmed Apr 06 21:58:39 that silent.sh sets profile to silent with: dbus-send --type=method_call --dest=com.nokia.profiled /com/nokia/profiled com.nokia.profiled.set_profile string:"silent" Apr 06 21:59:00 ok Apr 06 21:59:26 iirc there may be issues depending on which user runs the script Apr 06 21:59:37 wondering why it says cant find my script.. Apr 06 21:59:44 ok Apr 06 22:00:16 same also with user and root Apr 06 22:00:21 sh silent.sh works Apr 06 22:00:55 either way, phone-control sends the same command, and 'always' works :) Apr 06 22:03:18 hmm Apr 06 22:03:24 but this worked previously Apr 06 22:03:39 had to reflash phone so had to reset things Apr 06 22:03:47 and I can't remember anymore :D: Apr 06 23:30:04 we've seen this before Apr 06 23:30:32 iirc something with shebang line in script Apr 06 23:31:07 check if or if not you got a space behinf #! Apr 06 23:41:00 hmm, nope. can't reproduce problem like that Apr 06 23:42:04 aaah wait. wasn't that a msg that meant "can't execute the interpreter mentioned in shebang"? Apr 06 23:42:13 DocScrutinizer: it worked Apr 06 23:42:34 what worked? Apr 06 23:42:39 but doesn't still exec in fcrontab Apr 06 23:42:56 correcting /bin/sh Apr 06 23:43:07 hah, knew it Apr 06 23:43:13 ;p Apr 06 23:43:14 extra space? Apr 06 23:43:19 thx Apr 06 23:43:32 had bash there :p Apr 06 23:43:36 yeah Apr 06 23:44:08 but still fcrontab doesn't work Apr 06 23:44:28 I dunno which user runs fcrontab Apr 06 23:44:57 typically you need run-standalone.sh for things like cronjobs Apr 06 23:45:33 teid with root, yea Apr 06 23:45:36 * tried Apr 06 23:45:53 /usr/bin/run-standalone /home/user/silent.sh Apr 06 23:46:48 QUIT Apr 06 23:47:17 maybe you have replaced your standard shell? Apr 06 23:47:42 btw it's run-standalone.sh Apr 06 23:47:53 sec Apr 06 23:48:00 yea :p Apr 06 23:48:09 that it was ;p Apr 06 23:48:56 DocScrutinizer: thx! Apr 06 23:49:02 YW Apr 06 23:49:14 :) Apr 06 23:49:16 works now Apr 06 23:49:29 nice to have automated profiles Apr 06 23:49:59 cos dont wanna wakeup when some ones calls at 4 am :) Apr 06 23:50:38 well, when somebody calls me at 4 am, I usually better wake up ;-D Apr 06 23:50:48 :D Apr 06 23:51:01 cos of work? :p Apr 06 23:51:21 dunno, wuldn't know when I got a call at that time Apr 06 23:52:13 must've been 10 years ago, when a friend was terribly desperate Apr 06 23:53:58 anyway, I think I should look into dwimd again. Now I actually need it to set silent profile when at work Apr 06 23:55:35 the point is I'd prefer this not to be bound to a fixed time of day Apr 07 00:02:42 ok :) Apr 07 00:57:30 Happy easter and other bla-bla. Apr 07 00:58:01 I'm sure no one is reading this for mentioned reason, yet in case of other pagans like me sitting here: Apr 07 00:58:07 * DocScrutinizer wants some chocolate eggs! Apr 07 00:58:28 does anyone know which gstreamer0.10-plugins one should install for, erm, best result. Apr 07 00:58:44 * Estel_ handles some chocolate eggs to DocScrutinizer, sick of them already Apr 07 00:58:51 define "best results" ;-D Apr 07 00:59:00 I was just to do so :D Apr 07 00:59:26 by best result, I mean most widely working playback including strange format Apr 07 00:59:39 + most effective decoding (for non-DSP accelerated) Apr 07 00:59:44 all? Apr 07 00:59:53 ;-) Apr 07 00:59:55 not possible, they're exclusive Apr 07 01:00:00 aaah Apr 07 01:00:05 amongst conflicting ones, we have: Apr 07 01:00:13 gstreamer0.10-plugins-bad Apr 07 01:00:18 gstreamer0.10-plugins-bad-multiverse Apr 07 01:00:28 gstreamer0.10-plugins-ugly Apr 07 01:00:32 gstreamer0.10-plugins-ugly-multiverse Apr 07 01:00:32 well, then you have to decide. I guess the tradeoff is between stability and wide support for strange formats Apr 07 01:00:45 same thougths here... Apr 07 01:00:53 I've used "bad" for ages... Apr 07 01:01:05 Yet, thinking about it, I have NFC what ugly is exactly Apr 07 01:01:10 google might help Apr 07 01:01:15 "plugins that could pose distribution problems" Apr 07 01:01:20 tried that :P Apr 07 01:01:30 meh, who cares distri problems? Apr 07 01:01:30 wtf is that, sounds like some cop,yright shit Apr 07 01:01:38 yet, theyre distributed Apr 07 01:01:54 wasn't able to figure which one is more bleeding-edge - ugly or bad Apr 07 01:01:58 (sic!) Apr 07 01:02:01 well, they're not *installed* Apr 07 01:02:08 also, wtf is multiverse variant. Apr 07 01:02:17 the good, the bad and the ugly Apr 07 01:02:22 yea Apr 07 01:02:25 How to export conversations of N900? Apr 07 01:02:32 fortunatelly, good isn't exlusive to bad here Apr 07 01:02:55 LaoLang_cool: dunno if there's any serious method Apr 07 01:03:07 the problem is - multiverse, as it sounds ubuntu'ish, is "more" or less than non-multiverse variant (for Maemo) Apr 07 01:03:31 hmm, check content? Apr 07 01:03:40 LaoLang_cool, search repos for "conversations", there is something that exports it from database Apr 07 01:04:01 hm, why not, but how do I determine which one is "better" Apr 07 01:04:05 number of plugins? Apr 07 01:04:11 dates? ;P Apr 07 01:04:35 I'm also afraid that multiverse wariants got same plugins compiled differently Apr 07 01:04:44 or from diferrent code revisions Apr 07 01:04:46 headache Apr 07 01:04:58 * LaoLang_cool searched in http://maemo.org/packages/search/ but without lucky... Apr 07 01:05:44 dates, versions, number of plugins Apr 07 01:06:24 yeah, quite some headache Apr 07 01:06:44 LaoLang_cool, check glogarchive Apr 07 01:06:48 anyway listen to LaoLang_cool - might be a starting point for your question Apr 07 01:07:42 Estel_: wow, will try it, thanks Apr 07 01:07:47 DocScrutinizer: what?... Apr 07 01:07:56 "search, view, archive and export SMS/IM messages and call logs from eventlogger" Apr 07 01:09:11 LaoLang_cool, I got quite unsaatisfying results when searching via packages/search, so I always use fapman for repo search (with devel enabled, usual warnings apply) Apr 07 01:09:30 it search not only for package names, but crawls also through descriptions Apr 07 01:09:51 DocScrutinizer, thanks, will try to determine that. Now I remember we was talking about it some months ago... Apr 07 01:10:09 When I wanted to play .vob file with mpeg2 stream via stock media player :P Apr 07 01:10:42 this is why I installed "bad" plugins set on first place. Well, .vob still plays only via mplayer... Apr 07 01:10:55 GStreamer Bad Plug-ins is a set of plug-ins that aren't up to par compared to the rest. They might be close to being good quality, but they're missing something - be it a good code review, some documentation, a set of tests, a real live maintainer, or some actual wide use Apr 07 01:11:09 vs Apr 07 01:11:13 This packages contains plugins from the "ugly" set, a set of Apr 07 01:11:15 good-quality plug-ins that might pose distribution problems. Apr 07 01:11:35 now I'm trying to find most efficient theora playback. It's really good already, but for some videos I can't play smoothly, despite encoding via same settings as working ones Apr 07 01:11:36 yea Apr 07 01:11:54 from common sense "bad" sounds more bleeding-edge = wider format support Apr 07 01:12:07 yet, ugly description is enigmatic, to say at least Apr 07 01:12:23 not really Apr 07 01:12:36 I read that as plain "M$ codecs" Apr 07 01:12:43 ? so it's just me who don't know what distribution problems mean Apr 07 01:12:49 yes, same here, but Apr 07 01:12:57 they're distributed anyways via repos Apr 07 01:13:06 not installed though Apr 07 01:13:08 where illegal content is fought with fire, yep? Apr 07 01:13:15 hm? Apr 07 01:13:25 define "not installed" in this content Apr 07 01:13:34 Nokia neither supports nor ships them Apr 07 01:13:39 context* Apr 07 01:13:39 ergo no problem Apr 07 01:13:57 extras repo is not nokia repo Apr 07 01:14:03 yea, but this sucker conflicts with "bad" set Apr 07 01:14:16 yes, but bad is also in extras, not nokia repos Apr 07 01:14:26 eh, manual determining wtf is inside is a must :/ Apr 07 01:14:27 sure Apr 07 01:14:41 why ugly can't coexist wikth "bad" though Apr 07 01:15:04 I understand it as base < good < bad Apr 07 01:15:16 base is everywhere, good (with extras) can be installed Apr 07 01:15:26 bad (with extra variant) too Apr 07 01:15:35 none conflict with eachother Apr 07 01:16:04 with exception for bad replacing good, as it provides it already Apr 07 01:16:19 yet, checking dependencies, ugly doesnt provide bad, yet conflicts with it and replaces Apr 07 01:16:24 bad also replace ugly Apr 07 01:16:27 btw beware, I seem to recall this whole mess gave me more headache: after installing one, some codecs didn't work anymore Apr 07 01:16:56 that's what I'm afraid of, and why I asked here instead of determining myself via experiments :P Apr 07 01:17:27 add to this multiverse variants of bad and ugly (good also have multiverse variant, but that doesn't matter here...)) Apr 07 01:17:34 backupmenu is your friend ;-D Apr 07 01:17:39 true :P Apr 07 01:18:07 hah, worst case scenario is "no distinguishable difference between them", until You ran into problems 6 months later :P Apr 07 01:18:30 plus I'd forget about multiverse versions, I can't even seem to find them in package interface Apr 07 01:18:44 ooh there Apr 07 01:20:06 at least multiverse variant wasn't uploaded by same person Apr 07 01:20:12 another question, any tools for replacing battery icon in status with battery level, so I can check how many batter existed when click the status bar Apr 07 01:20:15 anyway multiverse is 0.10.7 while non-muperverse is 0.10.14? Apr 07 01:20:24 yet, dates for every gstreamer0.10 package is unavailable Apr 07 01:20:26 hm Apr 07 01:20:34 yyy, no? Apr 07 01:21:15 0.10.10.2-1maemo2 both variants for ugly Apr 07 01:21:37 oops 0.10.10 right Apr 07 01:22:24 multiverse was uploaded by solme guy with @ubuntu.com Apr 07 01:22:33 Repository Latest version Apr 07 01:22:35 Fremantle Extras-devel free armel gstreamer0.10-plugins-ugly-multiverse 0.10.7-2maemo Apr 07 01:22:38 hm, maybe only difference is that it provided LAME mp3 encoder? Apr 07 01:22:52 woot? Apr 07 01:23:33 oh god Apr 07 01:23:53 multiverse variants haven't dependencies set to conflict with bad and good Apr 07 01:23:59 i.e. normal ugly replaces them Apr 07 01:24:05 yet, multiverse doesn't Apr 07 01:24:16 it may be why You experienced codec breaking Apr 07 01:24:53 also, multiverse ugly variant have set that it provided LAME, while ugly doesn't (which doesn't mean that it isn't providing it, may have just dependencies unset) Apr 07 01:25:17 I pray for ressurection of some gstreamer/codec guru to clean this mess Apr 07 01:25:52 who the fuck have now idea of what exactly is bundled there :P Apr 07 01:26:10 bad-extra, bad-faad, bad-vorbis ;-P Apr 07 01:27:03 I'd guess maybe the dependecies are a bit messed up Apr 07 01:27:18 probably yes Apr 07 01:27:24 BTW I was wrong it seems Apr 07 01:27:33 only "ugly" have multiverse wariant Apr 07 01:27:37 maybe dpkg -install -force-install helps, when doen in right sequence? Apr 07 01:27:49 quite possible Apr 07 01:28:24 BTW there was once almost finished implementation for DSP decoding of Theora Apr 07 01:28:43 hmm Apr 07 01:28:44 actual playback on N900 was achieved with great results, it needed only some code polishing Apr 07 01:28:54 havent heard about it for ages, which is huge pity Apr 07 01:28:56 and now? Apr 07 01:29:10 especially that it was actual working as it seems Apr 07 01:29:43 http://talk.maemo.org/showthread.php?t=50270 Apr 07 01:29:50 no friggin maintainers Apr 07 01:29:52 see article linked on first post Apr 07 01:30:10 yea. Google distributed financial grant for theoraarm Apr 07 01:30:23 =version just after grant was last one Apr 07 01:30:43 ironically, before grant, work was fast and non-granted Apr 07 01:30:55 (for theoraARM) Apr 07 01:31:06 N900 implementation was based on that Apr 07 01:31:26 it was sheduled for mainline inclusion in theora and then... Apr 07 01:31:34 I have no idea what happened Apr 07 01:32:20 yet, even using already existing implementation, article's author from link I've provided was able to decode 800x480 theora @ ~30 fps with 80% idle CPU Apr 07 01:32:44 given pulseaudio eating things etc, 60% idle non-overclocked cpu seems real Apr 07 01:33:38 now, @ 900 mhz I got 80-90%used CPU for fluent playback for most videos, yet, some of them - same resolution (720x480, dvd rips) and same quality use 100% CPU and stutter a little Apr 07 01:34:00 no idea what's the difference, probably upper limit of kbps handled by CPU alone. Apr 07 01:34:22 (more dynamic scenes or whatsnot = more kbps using same resolution and same quality setting) Apr 07 01:34:50 * ShadowJK wonders how ffmpeg/libav's theora decoder on arm does vs xiph and theoraARM Apr 07 01:35:28 yea, quite interesting. Yet, in latest ffmepg releases I wasn't able to find theora decoder Apr 07 01:35:33 either I'm just dumb Apr 07 01:35:41 or it was removed after some version Apr 07 01:36:27 BTW, I would be happy to have either one actually available for N900 :P now, I can probably go and compile one myself, if desperate... Apr 07 01:36:44 it's probably in the same files as the vp3 decoder Apr 07 01:36:53 whichp given my miserably skills, is far from promising optimal results:P Apr 07 01:37:09 w00t? Apr 07 01:37:15 * Estel_ is running to check that Apr 07 01:38:03 yep, seems it's in vp3 Apr 07 01:38:08 same thing basically Apr 07 01:38:45 haha, so that's why it 'vanished'? :-D Apr 07 01:38:59 it seems so. Apr 07 01:39:19 yet, ffmep also provided encoder for it (not on topic about N900, of course) Apr 07 01:39:30 now I can't find encoder for theora or vp3 either Apr 07 01:39:32 suckers forgot a "symlink" somewhere in the wiki Apr 07 01:39:37 decoder is there, right Apr 07 01:39:42 ya. Apr 07 01:40:19 BTW I was forced to use ffmepg2theora standalone, instead of ffmpeg, when doing my re-encoding into theora Apr 07 01:40:29 due to "vanished" encoder. Apr 07 01:40:39 wasn't able to google "why" either. Apr 07 01:40:53 ffmpeg never had an encoder Apr 07 01:41:06 it had, I have old version with it on hd Apr 07 01:41:13 will check number Apr 07 01:41:23 At some point it was maybe able to use xiph's libtheora, but that wrapper was probably dropped as soon as the vp3 decoder could handle decoding the theora variant too Apr 07 01:41:26 via libtheora, if I'm not mistaken Apr 07 01:41:41 yea Apr 07 01:41:46 exactly Apr 07 01:42:06 now I understand... Dit I mention my miserabl.e knowledge on first place?:P Apr 07 01:42:13 who's going to ship me some chocolate eggs and a bottle of rum? Apr 07 01:42:46 BTW, ShadowJK, so how one can now encode something to theora via ffmpeg? just can't? Apr 07 01:42:57 correct Apr 07 01:43:24 * ShadowJK wonders if x264 ever made a vp8/webm version Apr 07 01:43:56 for the decoder in libav, the patch that added WebM support was 1000 lines of code (including comments) against h264. :P Apr 07 01:44:15 strange, format is becoming more and more competetive in terms of video quality and file size, yet, to encode anything one must compile or use ffmpeg2theora Apr 07 01:44:18 heh Apr 07 01:45:11 It's a dead-end Apr 07 01:45:12 ShadowJK: what's your favourite selection of codecs now, for fremantle? Apr 07 01:45:23 BTW, any (averagely) easy way to check ffmpeg arm theora decoder and/or theoraARM on-device? Apr 07 01:45:31 uh?! h264 baseline profile, it's the only sensible choice Apr 07 01:45:41 or one must re-do work and compile results? Apr 07 01:45:45 hm Apr 07 01:46:11 oh, iirc, ffmpeg/libav's vorbis decoder had superior battery life in tkulve's tests too :) Apr 07 01:46:12 gstreamer-plugins Apr 07 01:46:25 I'm not on-topic for codecs now - why it's dead-end? Apr 07 01:46:30 yes Apr 07 01:46:56 ShadowJK: FFmpeg was certainly able to encode to Theora. Apr 07 01:46:57 but I would like to be able to (quite boring-mainstream'ish) already use that :P Apr 07 01:47:04 i'm itchy about friggin mediathek at 3sat.de not working anymore Apr 07 01:47:07 (using libtheora) Apr 07 01:47:39 ShadowJK, any way to practically use results of tkulve's tests? Apr 07 01:48:17 having to choose between no implementation at all or unpolished implementation, I would preffer the latter Apr 07 01:48:20 Estel_, It's like a mpeg2 codec, that's been tweaked in the details to be different from mpeg2, and there's only so much gain you can get from tweaking the encoder for that. At some point if you want to keep up you want to get all the new stuff you can do in video coding now that CPUs are much more powerful than that of 15 years ago Apr 07 01:49:10 I don't think he released any debs to use/override/prefer libav vorbis decoder over other decoders. The libav vorbis decoder is quite mature anyways. Apr 07 01:49:11 I see. Apr 07 01:49:35 yea, I'm quite sure no debs exist now Apr 07 01:49:42 http://www.3sat.de/mediathek/index.php?display=1&mode=verpasst&type=1&red=nano >.-( Apr 07 01:49:49 but, maybe something at least compiled for manual replace? :P Apr 07 01:50:17 or one need to setup dev enviromnent and compile? Apr 07 01:50:43 BTW, so, You think that Theora doesn't have chance in long run, despite all things about being patent-free? Apr 07 01:51:05 There doesn't exist any patent free software Apr 07 01:51:22 all software is patent-free. Apr 07 01:51:35 all after all, mp3 hegemony was also present (is?), yet ogg kick ass in terms of practical quality, which interest me most... Apr 07 01:51:59 luke-jr, tell that to someone in USA (not me, fortunately) Apr 07 01:52:03 Estel_: Anyway, theoraarm and the DSP port were two different things. Apr 07 01:52:13 hm? Apr 07 01:52:14 Estel_: patenting algorithms isn't valid in the US either. Apr 07 01:52:31 The latter was called Leonora, and ran on the C64x DSP, not the ARM chip. Apr 07 01:52:53 agree with You meritocraticelly, (luke-jr), yet, practice seems to be populistic'ish different :( Apr 07 01:52:57 theorarm was just providing ARM asm optimizations. Apr 07 01:53:18 Both projects were simply folded into mainline libtheora. Apr 07 01:53:25 wait, they got 80% idle 600 mhz CPU due to small optimizations? Apr 07 01:53:41 kulve was mentioning dsp for sure Apr 07 01:53:55 TI DSP, exactly one we have here Apr 07 01:54:09 HAH, mediaplayer format works again :-D Apr 07 01:54:10 Yes. The DSP code was Leonora. Apr 07 01:54:15 I helped write most of this code. Apr 07 01:54:23 also, I have 80% used cpu when playing back theora? why? Apr 07 01:54:29 excuse my ignorance here. Apr 07 01:55:22 ah, I get it Apr 07 01:55:29 leonora = dsp Apr 07 01:55:38 c64x dsp = our dsp Apr 07 01:55:44 sorry, was lost for a while Apr 07 01:55:56 Well, welcome back, then :). Apr 07 01:56:02 so, if boths projects are mainline now, why don't we benefit? Apr 07 01:56:54 playing back thneora on N900 is using cpu only for sure (counting by CPU usage) Apr 07 01:57:15 Well, the DSP code still has some limitations that were never solved. Apr 07 01:57:25 no idea about CPU optimizations, but dsp seems to sleep. Apr 07 01:57:39 You still need to run a special version of the library on the host CPU that passes things through to the version of the library running on the DSP. Apr 07 01:57:51 I see. So, it wasn't implemented here, in our codec set? Apr 07 01:57:55 There are some issues with cache coherency that I think are just kernel bugs. Apr 07 01:58:12 maybe kernel-power team would be able to fix that? Apr 07 01:58:15 (e.g., decoding videos at QCIF resolutions is just broken) Apr 07 01:58:17 derf: so fix them in powerkernel? Apr 07 01:59:07 DocScrutinizer: If I thought the N900 was still worth the time investment as a development platform, sure. Apr 07 01:59:15 "team" :-D. nice euphemism, Estel_ Apr 07 01:59:15 also, http://blog.mjg.im/2010/04/16/theora-on-n900.html mention actual comparisions in performance, so bugs weren't critical? Apr 07 01:59:32 DocScrutinizer, 2 people is a company already, no? :P Apr 07 02:00:11 derf, ok, what You would recommend as way to go for people that think N900 is worth the time investment? Apr 07 02:00:22 Estel_: The kernel bugs? As I said, they were cache coherency things, and only affected decodes at small resolutions (i.e., when an entire frame fits in L2 cache). Apr 07 02:00:38 1. put what we have already to -devel 2. fix kernel bugs 3 code cleanup Apr 07 02:00:43 4. new version? Apr 07 02:00:58 Personally I'd use a libtheora build from current SVN. Apr 07 02:01:12 And forget the DSP stuff. Apr 07 02:01:25 why? Apr 07 02:01:59 Theora as is (without dsp) seems to be quite updated on N900 Apr 07 02:02:26 I know that cpu is still able to decode more effectively than dsp (they can't cooperate 100%, yes?) Apr 07 02:02:31 but battery savings? Apr 07 02:02:55 what makes you think there are battery savings? Apr 07 02:02:57 using dsp-assisted decoding and ~20% CPU used I got ~210 mA current Apr 07 02:03:22 DSP needs power too Apr 07 02:03:25 decoding theora @ 80-90% CPU, same other settings = ~410 mA Apr 07 02:03:39 yea, but it's roughly 100% more Apr 07 02:03:44 hmm Apr 07 02:03:52 maybe, maybe not Apr 07 02:04:09 with dsp thing CPU handles screen and some sound things (pulseaudio main eater) Apr 07 02:04:21 you'd need to compare same algo, same media, same backlight settings and all Apr 07 02:04:39 same backlight settings were compared, ambient light etc. Apr 07 02:04:52 of course I agree for algo etc Apr 07 02:04:52 It's actually a bunch of work to get the stuff from the DSP to the screen. Apr 07 02:05:05 yep, xorg handles that, no? Apr 07 02:05:17 uh Apr 07 02:05:27 The stuff that was done for Leonora was based on bc-cat, because the ultimate goal was to use it in Firefox, which can't use the hardware overlays. Apr 07 02:05:30 I imagine it's a bunch of work to get stuff from the DSP out of the dsp Apr 07 02:05:40 Unfortunately, the TI bc-cat driver was bugged to hell. Apr 07 02:05:54 It would usually fail to work the first time you tried to play a video. Apr 07 02:06:06 And then it would randomly reboot your device other times. Apr 07 02:06:13 Hm Apr 07 02:06:21 Not using overlay gobbles up lots of CPU and bandwidth :/ Apr 07 02:06:59 ShadowJK: Well, that was the point of using bc-cat, which was supposed to let you feed things to GPU as OpenGL textures directly. Apr 07 02:07:12 And indeed it did, when it worked. Apr 07 02:07:37 But since you're not Firefox, you'd probably want to use the hardware overlay instead. Apr 07 02:08:00 overlay used 3-4x more CPU; yet allowed more frames per second Apr 07 02:08:14 3-4x more CPU than what? Apr 07 02:08:19 OpenGL/bccat was resulting in lower fps, yet almost idle CPU Apr 07 02:08:22 the idea kinda freaks me out, since this is only my second PC that has a GPU fast enough to render a video texture :-) Apr 07 02:08:24 than the latter Apr 07 02:08:32 accordsing to kulve's article Apr 07 02:09:19 (video decode + paint, audio decode + not played) Apr 07 02:09:44 SDL/overlay - 800x480 - fps 33 ṔU 80% usage Apr 07 02:09:49 Estel_: Which article is this? Apr 07 02:10:11 OpenGL/bc-cat - 800x480 - fps 26 CPU 20%usage Apr 07 02:10:18 PUwas meant to be CPU Apr 07 02:10:30 You mean http://blog.mjg.im/2010/04/16/theora-on-n900.html ? Apr 07 02:10:39 That was written by Matthew Gregan. Apr 07 02:10:44 Not kulve. Apr 07 02:10:54 http://blog.mjg.im/2010/04/16/theora-on-n900.html Apr 07 02:10:59 ah sorry Apr 07 02:11:04 (Matthew Gregan is a Firefox developer) Apr 07 02:11:06 true. Apr 07 02:11:41 ok, but if we screw firefox and normal video playback (most likely, via open media player = stock player replacement) interest us Apr 07 02:11:52 no way to squeeze something useful from leonora? Apr 07 02:12:15 considering possibility to fix kernel bugs Apr 07 02:12:29 bc-cat is a dead-end? Apr 07 02:12:34 it's that bad written? Apr 07 02:13:08 I mean, if you want to spend the time to track down kernel bugs and try to get bc-cat issues fixed, then things could probably be made to work okay. Apr 07 02:13:27 it's probably crowded with arcane workarounds for silicon bugs Apr 07 02:13:49 kernel bugs can be probably traced down, we got brillant kernel developers now. I'm more worried about bc-cat Apr 07 02:13:57 for reasons DocScrutinizer mentioned :P Apr 07 02:14:09 and Your info about "buggy as hell" Apr 07 02:14:57 BTW, on slightly different matter - DocScrutinizer, do You imagine a sensible way for - once for all - determining max bandwidth hostmode is giving us? Apr 07 02:15:04 I'm still worried about sulu's result Apr 07 02:15:27 would like to know if it's really that limited, and if yes, why? kernel hostmode implementation bugs? Apr 07 02:15:36 I'd not be surprised if hostmode was slow as geotectonics Apr 07 02:15:42 more and more very interesting things to do with hostmode appear Apr 07 02:15:53 Estel_: Well, it looks like bc-cat has a much newer version than when we were trying this stuff two years ago. Apr 07 02:15:58 So some of those bugs may have been fixed. Apr 07 02:16:12 I'm especially interested in already working multi-track sound recording via hostmode Apr 07 02:16:15 hm Apr 07 02:16:22 derf, interesting Apr 07 02:16:34 In any case it has _some_ source code. Apr 07 02:16:37 after all we've tweaked the kernel quite nasty to get it to do hostmode at all Apr 07 02:16:43 (links from http://processors.wiki.ti.com/index.php/OpenGLES_Texture_Streaming_-_bc-cat_User_Guide) Apr 07 02:16:59 So maybe some of those bugs are fixable. Apr 07 02:17:05 and nobody ever looked into all the clocks and powersavings mode settings for the different busses etc Apr 07 02:17:58 derf, now I wait for RSA system-shock like skills learning, and I jump into coding right now ;) Apr 07 02:18:18 seriously though, huge thanks for dropping in with such valuable info, I'll check what can be gained from that Apr 07 02:18:49 and if we have manpower to do bc-cat thing, without that no much sense in bugging kp people Apr 07 02:19:14 I'd try to stream something via dd from /dev/zero to a sink in a gadget that takes unlimited bandwidth Apr 07 02:19:18 DocScrutinizer, maybe it's good time to do tests and start tweaking,when we have active kernel-power people Apr 07 02:19:41 I know usb-kernel is narrow specialization Apr 07 02:19:48 indeed Apr 07 02:20:09 but, maybe middle-way we can make our supermans specialized :P Apr 07 02:20:12 dm8tbr might be able to look into it, maybe Apr 07 02:20:30 neither pali nor freemang have the knowhow Apr 07 02:20:43 it would be pity to waste such potential, 5.1 playback from N900, and, more importantly, recording... Apr 07 02:20:59 hm? Apr 07 02:21:13 wtf is dm8tbr? I'm probably sleepy Apr 07 02:22:08 and, for mortals - what type of gadget You have in mind Apr 07 02:22:25 nothing special yet, still pondering Apr 07 02:22:38 /whois is your friend Apr 07 02:22:43 and how to measure speed of streaming this dd from dev/zero to sink in our gadget. with stopper? :P Apr 07 02:23:04 dd does that for you Apr 07 02:23:14 right. I'm sleepy indeed Apr 07 02:23:29 usb sound card doesn't seem to accept unlimited bandwidth, eh? Apr 07 02:23:36 or does it? Apr 07 02:23:59 virtually unlimited, given you can set sampling rate and num of chan high enough Apr 07 02:24:46 erm, but dd'ing something to sink there? Apr 07 02:24:53 any noob friendly command example? Apr 07 02:24:55 err, yep Apr 07 02:25:00 right sound odd Apr 07 02:25:26 aplay might do Apr 07 02:25:26 definitely, yet, very interesting at the same time. Haven't done wuch thing. Apr 07 02:26:00 aplay on N900 side or dd to aplay? /noob mode Apr 07 02:26:01 aplay -r 512000 -c 6 -f s48le Apr 07 02:26:08 oh, nice, thanks Apr 07 02:26:42 maybe there's actually s32le Apr 07 02:26:45 format Apr 07 02:27:04 and max sampling -rate is card dependant Apr 07 02:27:17 as is number of -chan Apr 07 02:27:33 Recognized sample formats are: S8 U8 S16_LE S16_BE U16_LE U16_BE S24_LE S24_BE U24_LE U24_BE S32_LE S32_BE U32_LE U32_BE FLOAT_LE FLOAT_BE FLOAT64_LE FLOAT64_BE IEC958_SUBFRAME_LE IEC958_SUBFRAME_BE MU_LAW A_LAW IMA_ADPCM MPEG GSM SPECIAL S24_3LE S24_3BE U24_3LE U24_3BE S20_3LE S20_3BE U20_3LE U20_3BE S18_3LE S18_3BE U18_3LE Apr 07 02:28:35 ok. so after that command, tweaked to my 24 bit 96khz card, to measure speed, I need to? Apr 07 02:29:06 float64_be :-D Apr 07 02:30:16 aplay -f s24_le -r 96000 -c $NUM-OF-CHANNEL-CARD-SUPPORTS Apr 07 02:30:54 it seems im not that dumb, figured that :P (audio command) Apr 07 02:31:17 DocScrutinizer: For when you need 12,000 dB of dynamic range. Apr 07 02:31:23 still don't get how to measure speed hostmode is providing Apr 07 02:31:38 derf: :-D Apr 07 02:31:45 derf, hahaha, we check how much bandwidth does hostmode actually provide :P Apr 07 02:33:22 so, for dumbie, alsay will play my 40khz 16bit file upsampled to 24bit 96khz via sound card. ok. how to check hostmode is allowing, again? Apr 07 02:33:39 also, I suppose 24bit 96khz is enough to check our speed limit? :P Apr 07 02:33:55 and wtf is float64_be Apr 07 02:34:42 le... be... the thing I'm thinking when I see those in combination is big/little endian... I suppose this is not really what it is about though Apr 07 02:35:40 ok, screw float64, how to actually check this damn bandwidth? Apr 07 02:35:43 well, 24bit * 96k * 4chan = 2Mbit Apr 07 02:35:52 FIQ: Yes, that is exactly what it is. Apr 07 02:35:52 *6 chann Apr 07 02:36:01 oh Apr 07 02:36:01 3Mbit ;-P Apr 07 02:36:03 ok Apr 07 02:36:08 well, card is able tp 7.1 actually, so I could use 8? Apr 07 02:36:17 FIQ thanks anyway :) Apr 07 02:36:34 float64_be is some sapling format alsa supports Apr 07 02:36:38 sure Apr 07 02:36:43 big endian was right Apr 07 02:37:07 I got 5.1 speakers connected, card should handle it anyway Apr 07 02:37:16 still to small to check max bandwidth Apr 07 02:37:24 4Mbits Apr 07 02:37:38 well, it's sufficient bw for your usecase ;-P Apr 07 02:37:54 true :P but would like to know limits of hostmode Apr 07 02:38:00 You know, volounteering ;) Apr 07 02:38:13 also important for handling hdmi output from N900 via raspberry pi Apr 07 02:38:26 get a USB stick, check max continuous write on PC, check on N900 Apr 07 02:38:30 it would be lame to notice that WiFi is our fastest connection to outside world Apr 07 02:38:57 yea, but this could be limited by limiting read on N900 Apr 07 02:39:07 eh? Apr 07 02:39:16 I/O conflicts, whatever Apr 07 02:39:20 dev/zero has very few limits Apr 07 02:39:27 ah, true Apr 07 02:39:42 damn, sulu did that and he got poor results, You've already seen that Apr 07 02:40:07 I though that there may be some absolutely 100%sure way to measure once for all Apr 07 02:40:29 usb networking and streaming from /dev/zero to /dev/null in another devikce? :P Apr 07 02:40:41 get better data then! what speed got selected, did it actually run at that speed, was the stick able to get beter rates at PC under same test? Apr 07 02:41:05 ok... Apr 07 02:41:43 of course you as well could use a mere network connection via USB to a gadget, and flood it with socat or sth Apr 07 02:42:09 you still have tcp stack overhead then Apr 07 02:42:46 but the question is what you actually want to investigate and know Apr 07 02:43:10 mere raw USB bw helps nuttin if the overhead kills your app Apr 07 02:43:47 of course how much bandwidth hostmode on N900 can actually provide, and, in longer run, if it can be tweaked to ressemble actual hi-speed standards more, via tweaking kernel implementation Apr 07 02:43:52 after all the bottleneck could be anywhere Apr 07 02:43:57 of course Apr 07 02:44:17 Estel_: that's a moot definition Apr 07 02:44:20 thats why I would like different method than writing to sd card from dd, to have same or different results Apr 07 02:44:48 for example, most demanding thning I can now imagine - HDMI output from N900 via raspberry Pi on arbitrary high resolution Apr 07 02:45:14 of course You dont need 30MB/s for that, yet Apr 07 02:45:15 you always are testing a full chain of systems, from origin to final sink Apr 07 02:45:26 it would be good to know how good/bad our current implementation is Apr 07 02:45:38 and to be able to tweak it if needed Apr 07 02:45:39 yes Apr 07 02:45:51 yesthats why I'm opting for testing with few very different methods Apr 07 02:46:12 = less probability of every system involved in different test to be lame :P Apr 07 02:46:17 so if your HDMI driver is the bottleneck, your raw USB thruput figures are meaningless Apr 07 02:46:32 as they were done with a void driver Apr 07 02:46:49 erm, hdmi done via raspberry pi, the latter got data via usb networking Apr 07 02:47:02 depending on what the driver does on USB API your results may vary vastly Apr 07 02:47:18 generally, knowing hostmode current limits would also help determining why some things work as should and other don't Apr 07 02:47:38 for example, DVB-T mpeg4 sticks Apr 07 02:47:55 meh, what's the problem with bw there? Apr 07 02:47:59 more and more support for them in linux and people would start to test them with N900 Apr 07 02:48:24 honestly, HDMI and raspberry Pi is my priority now. If sulu's results are valid... Apr 07 02:48:34 WiFi is faster for us than USB networking Apr 07 02:48:44 he got terrible results, as You remember Apr 07 02:48:45 sulu's results are meaningless for your usecase Apr 07 02:48:51 first approach Apr 07 02:50:29 honestly, don't ask "how much can it do?". better ask "can it do XY?" Apr 07 02:50:44 somehow right Apr 07 02:50:45 yet Apr 07 02:50:59 Could You please look on that: Apr 07 02:52:33 http://talk.maemo.org/showpost.php?p=1172479&postcount=58 Apr 07 02:52:56 all after all You haven't answered him, and I would like to determine that even if just by curiosity Apr 07 02:53:13 I agree that we need to determine if tasks a) b) could be done Apr 07 02:53:34 if you're unhappy with a cetain performance on a particular "XY", ask what's the bottleneck in this particular usecase. Don't nag me with "how fast can I drive?" without specifying road or car Apr 07 02:53:41 yet, You stated there some interesting things to check, and conversation went dead after, would love to continue investigating Apr 07 02:55:47 first, investigate details of the test done. I already suggested ^^^^ Apr 07 02:56:03 well, using 2.5'' hdd or writing files via optical dvd writer for someone, You can get hurt by bootlenect Apr 07 02:56:11 hey, he answered about C state Apr 07 02:57:16 and asked about how to check CPU<->musbcore and musbcore<-ULPI->PHY interface Apr 07 02:57:46 C states for 10 minutes, CPU was in C0 Apr 07 02:57:57 You havent checked his answer, have You? Apr 07 02:58:25 dd if=/dev/sda of=/dev/null - CPU in C0 Apr 07 02:58:49 would like to perform Your further suggestions, yet need to know how to check Apr 07 02:59:01 CPU<->musbcore and musbcore<-ULPI->PHY interface Apr 07 02:59:22 unless You now got better idea for debugging that **** ENDING LOGGING AT Sat Apr 07 02:59:58 2012