IRC log of #maemo-ssu for Thursday, 2013-03-21

*** nox- has joined #maemo-ssu00:07
keriomerlin1991: as the... 25%? of the users of extras-devel-light, i demand those three packages to be updated!00:28
keriomaybe i even have them installed!00:29
merlin1991I shall gather statistics and only follow your demands, if you indeed are 25% of the users ;)00:29
*** Vlad_on_the_road has quit IRC00:33
kerioalso, if i'm really 25% of the users, we need better advertisement00:33
*** Martix has joined #maemo-ssu01:01
*** zogg__ has quit IRC01:26
*** zogg__ has joined #maemo-ssu01:27
*** Martix has quit IRC01:35
*** Martix_ has joined #maemo-ssu01:35
*** Martix_ has quit IRC01:37
*** dhbiker has quit IRC01:38
*** M4rtinK has quit IRC01:39
*** M4rtinK has joined #maemo-ssu01:51
*** M4rtinK has quit IRC01:56
*** arcean has quit IRC01:56
*** Martix has joined #maemo-ssu02:06
*** xes has quit IRC02:18
*** Martix has quit IRC02:23
*** Martix has joined #maemo-ssu02:24
*** Martix has quit IRC02:30
*** xes has joined #maemo-ssu02:31
*** xes has quit IRC02:37
*** sunny_s has quit IRC02:40
*** sunny_s has joined #maemo-ssu02:43
*** unclouded has joined #maemo-ssu03:02
*** nox- has quit IRC03:46
*** Pali has quit IRC03:57
*** kolp_ has joined #maemo-ssu04:21
*** kolp has quit IRC04:24
*** kolp_ has quit IRC05:10
*** amiconn has quit IRC05:13
*** amiconn_ has joined #maemo-ssu05:13
*** amiconn_ is now known as amiconn05:14
*** futpib has quit IRC05:24
*** sunny_s has quit IRC05:28
*** sunny_s has joined #maemo-ssu05:30
*** DocScrutinizer05 has quit IRC06:03
*** DocScrutinizer05 has joined #maemo-ssu06:03
Estel_no plans for shuttding down (hiding) extras-devel light, to make people resync with main repos again?06:12
Estel_sound like fragmentation06:12
Estel_btw, 10+ packages got build already06:13
*** M13 has joined #maemo-ssu06:13
Estel_including [shameless advert] nicest and greatest, like new build of bnf [/shameless advert] :P06:14
*** wumpwoast has quit IRC06:53
*** wumpwoast has joined #maemo-ssu06:53
*** LaoLang_cool has joined #maemo-ssu07:38
*** sunny_s has quit IRC07:39
*** sunny_s has joined #maemo-ssu07:42
*** LaoLang_cool has quit IRC08:20
*** M4rtinK has joined #maemo-ssu09:31
*** sunny_s has quit IRC09:32
*** sunny_s has joined #maemo-ssu09:35
freemangordonguys, help :) http://talk.maemo.org/showpost.php?p=1330432&postcount=14309:49
*** dhbiker has joined #maemo-ssu10:09
*** M4rtinK has quit IRC10:30
*** Martix has joined #maemo-ssu10:32
Estel_kerio,  dpkg -r hald-addon-bme is enuff to avoid device shutdown, when using replacement bme?10:57
kerioi don't think so10:57
kerioyou have to restart hal10:57
kerioto unload hald-addon-bme10:57
Estel_osso-actdead also wants to harakiri itself10:59
Estel_when no hald-addon-bme10:59
Estel_I hope it doesn't bork reboot ;)10:59
Estel_erm, hal restart = hald restart?11:00
Estel_kerio ^11:00
kerioidk11:00
Estel_freemangordon,  you've said that new pixman, even with your changes, performs "almost" as good as old one11:06
Estel_what are benefits of new one, then?11:07
Estel_some new possibilities, some bugs fixed (you said no bugs hitting old one)?11:07
Estel_from different barrel - where maemo loads wifi things for the first time? I would like to affect it by iw reg set JP11:21
*** futpib has joined #maemo-ssu11:48
*** M13 has quit IRC11:57
*** sunny_s has quit IRC12:03
*** sunny_s has joined #maemo-ssu12:06
freemangordonEstel_: new pixman opens the door for new xserver (for example). Also microb can be re-compiled to use system pixman, now it uses built-in with NEON accel disabled12:07
freemangordonEstel_: bew pixman means we can try to upgrade cairo as well12:07
freemangordon*new12:07
Estel_freemangordon,  so answer seems simple12:13
Estel_upgrade ;)12:13
Estel_Am I qualified to thumb up there, or is it targeted at coders only (serious question, 0 irony)12:14
Estel_btw, just in case you've forget, I really count on that q9700 module for usb->ethernet adaptors in next kp :P12:15
freemangordonEstel_: well, If you have an argument, it is should not matter if you are developer or not12:19
freemangordons/is should/should/12:19
infobotfreemangordon meant: Estel_: well, If you have an argument, it should not matter if you are developer or not12:19
Estel_those things about what new pixman means, were it presented in that thread already?12:19
freemangordonhmm, no12:24
*** unclouded has quit IRC12:27
keriofreemangordon: are the changes to /proc/mounts somehow backportable? do you know?12:39
keriothe ones that make it closer to /etc/mtab12:40
*** sunny_s has quit IRC12:48
*** Pali has joined #maemo-ssu12:52
*** Martix has quit IRC13:05
*** sunny_s has joined #maemo-ssu13:07
*** Estel_ has quit IRC13:10
freemangordonkerio: no idea13:38
*** sunny_s has quit IRC13:59
*** Estel_ has joined #maemo-ssu14:02
*** zeq has joined #maemo-ssu14:08
*** sunny_s has joined #maemo-ssu14:15
*** futpib has quit IRC14:16
*** kolp has joined #maemo-ssu14:26
*** sunny_s has quit IRC14:35
*** lizardo has joined #maemo-ssu14:37
*** sunny_s has joined #maemo-ssu14:38
merlin1991Estel_: I wont shut down devel light, because I use it myself :D15:09
merlin1991I'll sync it soon though15:09
merlin1991also it does have a distinct advantage of a smaller packages file15:10
freemangordonmerlin1991: what's your opinion re pixman?15:12
merlin1991no regressions --> thumbs up, weird shit --> no can do, stuff still todo --> hey you're developer, get on it ;)15:12
merlin1991s/on/to/15:12
infobotmerlin1991 meant: no regressitos --> thumbs up, weird shit --> no can do, stuff still todo --> hey you're developer, get on it ;)15:12
merlin1991arf15:12
merlin1991more than one on in the sentence15:12
merlin1991I just hope we don't run into more problems with our ancient X15:13
merlin1991btw HEL airport is nice15:14
* merlin1991 currently sits there waiting for the next flight15:14
freemangordonmerlin1991: well, there ARE some 5% speed regressions on some of the benchmarks, but vast improvement on some other15:14
freemangordonmerlin1991: that's why I am undecided.15:14
merlin1991we could try throwing in some debug stuff to measure usage in the general case15:14
merlin1991over a week or so15:15
merlin1991and if the speedups hit the more used parts and vice versa I'd say perfect15:15
freemangordonmerlin1991: well, IIUC it is not that simple. How we can estimate the speedup in microb rendering if compiled against system pixman? (just an example)15:16
merlin1991IIUC?15:16
freemangordon~iiuc15:16
infobotiiuc is probably If I Understand Correctly. Well, that's my best understanding.15:16
freemangordonmerlin1991: also, look at the benchmarks15:17
merlin1991dunno if we look at the usage of pixman stuff, and the faster stuff is used way more often, then it should speed up the system in general15:17
merlin1991will do15:17
merlin1991atm crafting a tmo message15:17
freemangordonmerlin1991: the preformance difference is some 2-5 %, well enough in the error margin15:18
freemangordon(excluding some exotic I'll fix if it makes sense)15:18
merlin1991freemangordon: any idea why all cairo demos are faster on stock?15:28
freemangordonmerlin1991: I guess upstream has a couple of more checks that lead to that small performance loss15:29
freemangordonmerlin1991: but it is not that bad, I mean, I can swallow 3% performance loss15:30
freemangordonmerlin1991: also when compiled with 4.7.2, it looks different ;)15:31
merlin1991yea15:31
merlin1991but that's future talk ;)15:31
freemangordonyeah :D15:31
merlin1991also you'd have to compare that to stock on 4.7.2 ;)15:31
freemangordonmerlin1991: I am not sure it makes sense15:32
freemangordonas the point to introduce newer pixman is not the performance15:33
merlin1991well possibly you have even higher regressions when comparing 4.7.2 compiled stock to pixmen15:33
merlin1991yeah I know15:33
merlin1991but we also don't want to get the system to a grinding halt by adding "smallish" speed regressions all over the system15:33
freemangordon merlin1991:no, iirc upstream performs way better with 4.7.2, while stock does not benefit much15:33
freemangordonit is 4.2.1 which can;t deal well with upstream C code15:34
merlin1991makes me wonder is gcc at fault or the upstream code? ;)15:34
freemangordonboth15:34
merlin1991err how can the code be at fault?15:35
merlin1991weird gcc > 4.2.1 specific tweaks?15:35
freemangordonmerlin1991: hmm, no, it is just that the code is more comlicated so 4.2.1 can;t optimize it well15:35
freemangordonthat's my understaning15:35
freemangordon*compilcated15:35
freemangordonfuck :D15:36
merlin1991yep the word itself is complicated aswell ;)15:36
merlin1991I'd really try to gather *some* statistics about the more/less used functionality pixman exports as a library, if the regressions fall into not so much used areas we can go forward without loosing any sleep15:37
freemangordonnaah, it is the stupid remote desktop over 3G, sometimes it does not register keypresses15:37
merlin1991remote desktop over 3g? dafaq?15:37
merlin1991why don't you use quassel, some shell + irssi, znc ?15:38
freemangordonmerlin1991: wish you luck, just look at haow many different code paths there are in the benchmark15:38
freemangordonmerlin1991: to windoze XP?15:38
merlin1991no to your irc client15:38
merlin1991so you have the keyboard handling locally ;)15:39
freemangordonaah, but I need my PC, i am doing some coding while taking rest from my job ;)15:39
merlin1991you do can tab between your remote desktop window and a local irc client ;)15:40
merlin1991also WTF do you code on windoze xp15:40
merlin1991kerio: I just checked apt-mirror does not find the new packages, (me guesses foobar db), that will take some time to fix15:47
freemangordonmerlin1991: My home desktop pc OS is Win XP15:47
merlin1991yeah, how can you possibly get anything done on there? :D15:48
freemangordonso it does not mater what i code, I do it either inder wn or under VM15:48
keriomerlin1991: maybe i didn't make it clear enough, but i was sarcastic ;)15:48
merlin1991kerio: even though people lately scream at me I still care about maemo ;)15:49
freemangordonmerlin1991: which reminds me... when will be the next meeting?15:50
merlin1991well next thursday or later15:50
merlin1991bah hel homepage only has terminal, but not gates for departing flights on the webpage, now I have to stand up and go over to the list xD15:51
*** LauRoman has joined #maemo-ssu15:54
freemangordonoh, poor merlin1991, he has to walk :P16:04
merlin1991kerio: currently resyncing my apt-mirror16:05
*** Estel_ has quit IRC16:23
*** Estel_ has joined #maemo-ssu16:23
Estel_urgent qauestion, please16:23
Estel_how to deal with f****n, broken package, that fails on deinstall due to fckd postrm script, and no matter of force for dpkg -r --force-all makes a difference?16:24
Estel_can't install any new things, and I'm urgently preparing a device for someone who is leaving soon16:25
keriodelete the postrm script from /var/dpkg/info16:25
Estel_huh?16:25
kerioand then uninstall16:25
Estel_doesn't even have /var/dpkg16:25
Estel_on device16:25
*** Martix_ has joined #maemo-ssu16:25
kerioer16:25
keriohmm, /var/lib/dpkg?16:25
Estel_(thanks for trying to help)16:25
Estel_yes, many catalogs there16:26
Estel_no idea where to look?16:26
kerio/var/lib/dpkg/info/packagename.postrm16:26
keriodelete it16:26
Estel_LOL things for all packages here, but nothing for php5-common16:27
Estel_yet, when trying to uninstall it:16:27
Estel_ah shit16:27
Estel_lol16:27
Estel_looking from wrong device16:27
Estel_*.*16:27
Estel_kerio,  thank you very much16:28
Estel_worked as charm16:28
Estel_really, how I suffered many times without knowing that16:29
kerioyeah, not reading the documentation causes that, often16:29
freemangordon:D16:29
Estel_I wonder why it isn't first google result, everyone f**ks about force commands, which doesn't work in most cases16:29
Estel_:D16:29
Estel_agreed16:29
Estel_but I tried, really, it's not my fault (tm)16:30
freemangordonit was like that (tm)16:30
Estel_.)16:30
kerioah shit what happened to your eye16:30
Estel_popped up while writing "I even forgive you for asslicking, during giving help on formatting syntax to ban me (which haven't worked, anyway)"16:31
Estel_;)16:31
Estel_so I'm even more glad it (advice) worked this time so "no-problem'ly"16:32
Estel_so I'm even more glad it (advice) worked this time so "no-problem'ly"16:32
Estel_brb, need to finish with that device16:32
Estel_sorry for echo16:32
*** gregoa_ has joined #maemo-ssu16:57
*** gregoa has quit IRC16:57
merlin1991funky my laptop keeps telling me it can run for 1 more hour16:57
merlin1991regardless how long it ran in the meantime16:58
keriomerlin1991: it's like my homework17:09
kerioi'll do it tomorrow17:09
merlin1991hm homework17:09
merlin1991I shoudl read a full book17:09
keriowhich kind of book?17:09
merlin1991but I'm currently in finland, without the book :D17:09
merlin1991actually I even forgot the title, and I have no idea what it contains, but we have to have read it after easter17:10
keriofor what?17:10
*** dhbiker has quit IRC17:12
*** discopig has quit IRC17:13
*** Martix_ has quit IRC17:13
*** discopig has joined #maemo-ssu17:14
*** discopig has joined #maemo-ssu17:14
*** dhbiker has joined #maemo-ssu17:15
*** Vlad_on_the_road has joined #maemo-ssu17:18
*** M4rtinK has joined #maemo-ssu17:27
*** andre__ has quit IRC17:27
DocScrutinizer05dang, andre__ left17:39
*** andre__ has joined #maemo-ssu17:40
DocScrutinizer05you're in Finland??17:41
DocScrutinizer05merlin1991: ^^^17:41
merlin1991DocScrutinizer05: yep, crashing nokia headquaters and staying there long enough untill they hand out source code17:41
merlin1991nah visiting friends .)17:41
*** M4rtinK has quit IRC17:44
DocScrutinizer05the fist variant sounded way more fascinating17:49
DocScrutinizer05first*17:50
*** BCMM has joined #maemo-ssu18:04
Estel_I'm afraid no one left in Nokia HQ, that know what source code is18:26
Estel_freemangordon,  before starting N900, had you linux background?18:27
*** M13 has joined #maemo-ssu18:28
Estel_I'm curious, as I seem to remember vaguely, that answer is "no" or "not a much". I'm just to, finally, set u proper VM and scratchbox in it, and even patchwork a maemo dev env on my device (via easy debian, or dedicated partition and plain chroot...), and start some coding18:28
Estel_probably, porting some terminal/CLI tools first to get a handle of how maemo works18:29
Estel_now, doc I've seen are a mess, some reffer to pr1.1, some to diablo too...18:29
Estel_apart from generic coding knowledge, where should beginner look for proper documentation about maemoities (thing needed only for maemo to compile)18:30
*** NIN101 has joined #maemo-ssu18:44
*** Woody14619 has joined #maemo-ssu19:02
*** Woody14619 has quit IRC19:02
*** Woody14619 has joined #maemo-ssu19:02
*** futpib has joined #maemo-ssu19:03
*** sunny_s has quit IRC19:07
*** marmistrz has joined #maemo-ssu19:09
marmistrzmerlin1991: ping19:09
marmistrzfreemangordon: ping19:09
merlin1991marmistrz: pong19:10
marmistrzI noticed the things that changed about hildon-desktop and X-CSSU-Force-Landscape19:10
marmistrzin latest version19:10
*** sunny_s has joined #maemo-ssu19:10
merlin1991hit arcean over the hea19:10
merlin1991*head*19:10
merlin1991he's not here though19:10
merlin1991btw cssu version, or version in cssu-devel?19:11
marmistrzhe said it's something about the new "policy"19:11
marmistrzcssu testing19:11
marmistrz(esp. thumb)19:11
merlin1991we did talk about it and X-CSSU-Force-Landscape should work again19:11
merlin1991don't know it that is in cssu-devel already19:12
marmistrzok, thx19:12
marmistrzjust wanted to make sure with the CSSU HQ :_19:12
marmistrzs/:_/:)19:12
*** zeq has quit IRC19:13
merlin1991hm looking at gitorious I don't see a related change, I'll talk to him when he shows19:13
marmistrzok, thx19:15
*** marmistrz has quit IRC19:17
freemangordonmerlin1991: afaik the change is in both gitorious and caasu-devel19:20
freemangordon*cssu19:20
merlin1991maybe already on page 2 of the changes :)19:21
freemangordoncould be19:21
*** M4rtinK has joined #maemo-ssu19:24
*** nightsh_ is now known as nightsh19:45
freemangordonmerlin1991: still no access to cssu repo?19:58
freemangordonwe have way above critical mass of packages, nuclear explosion is imminent :D19:59
Estel_is there a way to lock phone application to portrait in cssu?20:00
Estel_it's rotating even when status applet lock is on, and it drives me mad20:01
freemangordonEstel_: you need latest and greatest HD and lock applet ;)20:01
Estel_you mean lock applet from devel? I hope it'll make it way to cssu-thumb soon enough20:01
Estel_thanks anyway :)20:02
freemangordonEstel_: it will make it to -thumb as usual - after CSSU-T is out :)20:02
Estel_most of the times I'm just about to hit "answer" in landscape, and I end up hitting "reject" in portrait20:02
Estel_yea, that was allusion that I hope for cssu-t release :p20:02
Estel_in whatever repo20:03
Estel_does tech staf works about import problem? last time I asked what doesnt work yet, they forget to even mention it20:03
freemangordonno idea20:04
Estel_...while they remembered about things like non-working maemo.org account (not garage or wiki one)20:04
Estel_heh, merlin1991, how long ago you complained to tech staff about import non working so cssu can't get fixed in repos?20:04
Estel_I hope it wont reveal itself to be like autobuilder - everyne told me that no one know how to fix it all the time, then, I have asked once, and jacekowski answered he know how it works :P20:05
Estel_it's just that no one asked before20:05
Estel_maybe it's same this time?20:05
Estel_3 days later we had autobuilder working, maybe this time my enfant terrible habbit of asking straight question will help again :P20:07
Estel_shit, 5 GB home becoming too low20:07
Estel_and I don't install everything I see20:07
Estel_from different barrel - does anyone know how to run grub installer for maemo?20:08
keriowat20:08
Estel_I have something epic for maemo pentesting, waiting for arrival in repos20:08
Estel_but part of it is installing grub to /dev/mmcblk0p120:08
keriogrub as in the grand unified bootloader?20:09
Estel_it doesnt affect normal usage, I have it in my device20:09
Estel_yea20:09
Estel_:P20:09
Estel_part of it20:09
keriowhich grub?20:09
Estel_at least20:09
keriofor x86?20:09
Estel_yep20:09
Estel_i mean it's going to be used with x8620:09
Estel_I made kon-boot work with maemo20:09
Estel_and even got author permission to put it into non-free extras20:10
keriocan't you install it in the same way as you'd install it on a flash drive?20:10
Estel_sure. From desktop pc. but I would like to avoid any dependency on other devices20:10
Estel_me wanna make extras package for it20:10
kerioship an image, i guess?20:10
kerioor just make update-grub or whatever work20:11
keriocertainly *that* part isn't arch-dependent20:11
Estel_from device, you mean?20:11
Estel_yes, it shouldn't be20:11
keriomaybe you'll have to tell it to do the thing for x8620:11
Estel_well, as with image, I would like to avoid nuking file content on mmcblk0p1 of users ;)20:12
kerioinstead of complaining because "what is this and i am an ARM"20:12
Estel_hehe20:12
kerio...can you even install a bootloader on a single partition?20:12
keriowhere does it go?20:12
Estel_generally, maemo mass-storage mode denied kon-boot for working due to being non-bootable, even if grub installed, but I've found a way around it...20:12
Estel_no idea, but it works :P20:12
Estel_later I'll find equivalent of what I need to prepare partition from suckin windoze script20:13
Estel_and we'll try to find "the least we need" for maemo counterpart for that20:13
Estel_result is, no more need for carrying that pesky pendrives20:14
Estel_just connect your n900 to any desktop on boot, and you may skip password on any account ;)20:14
*** zogg_ has joined #maemo-ssu20:14
Estel_we will end in endgadget with their jaws dropping :P20:15
Estel_(that was a joke, I have endgadget in ass, honestly)20:15
Estel_still, it must look supercool to ignorants - one connect n900 to desktop, invoke terminal command, and OS on desktop - any windows or linux boot in administrator mode, without asking for password20:16
*** zogg__ has quit IRC20:16
Estel_imagine android lusers face20:16
keriothey can do the same fucking thing dude20:17
Estel_shhh!20:17
Estel_I knowp but they don't20:17
Estel_especially on endgadget. there they doesn't know anything20:17
Estel_+ androids require reboot20:17
Estel_n900 doesn't :P20:17
kerioalso, why the hell are you using mmcblk0p120:17
kerioinstead of a file20:17
Estel_because they mass storage won't be bootable while system is loaded20:18
Estel_because it don't hurt my mmcblk0p1 at all, loopback file might work too, if it would want to make itslf bootable20:18
kerioit won't help your mmcblk0p1 either20:18
Estel_in which way?20:18
kerioand with an image, you can also provide a bootable linux system for cheap20:18
Estel_thats true, too20:19
Estel_but as for mmcblk0p1, I haven't noticed any problems20:19
keriothere's no need to dick around with the user's data if you can just modprobe g_file_storage luns=1 bootable=1 file=whatever20:19
kerio(modulo correct parameters' names)20:19
Estel_but n900 as seed for installing linux distro in any desktop is cool idea that I've thought about before, focues on kon-boot for start20:19
Estel_ro=1 too20:20
Estel_hm, is it that simple?20:20
Estel_no need to fck with loopback images externally?20:20
Estel_nice20:20
Estel_ro=1, because g_file_storage doesn't unmount from maemo20:20
Estel_you DON'T want to get concurrent writes from maemo and mass storage20:21
kerioi think that writing to the file would work, too20:21
Estel_at the same time20:21
keriowhy not? it's just writing to a file20:21
keriomaemo isn't going to touch it, anyway, it's just an image20:21
Estel_sure, both OS'es not aware of what another one is doing20:21
Estel_try concurrent write to same thing20:21
kerio...you're not aware of what you're doing in this case, then20:21
Estel_i mean whole device20:21
kerioit's *a file*20:22
Estel_image file sits on some partition20:22
kerioshit writes to the same file all the time20:22
Estel_concurent write to same partition by two unaware oses could be catastrophic for that partition20:22
kerioyes, you might get corruption if there's no locking20:22
keriobut why would you write to that file anyway, from maemo?20:22
Estel_but not from 2 separate oses not knowing aboyt other existence20:22
Estel_that partition20:22
Estel_file sits on partition20:22
kerioyes, the filesystem drivers would be confused20:22
kerioIF THERE WERE MORE THAN ONE20:22
Estel_no idea how emmc hardware flash wearer would act, too20:23
keriounless you're going to also mount the same image file from maemo at the same time for some weird reason20:23
Estel_they made mass storage unmount for a purpose20:23
kerioyep, but you're going to use **a file**20:23
keriowhich isn't opened by anything on maemo's side20:24
Estel_but that file is behind partition which is behind single flash chip which is behind single controller20:24
Estel_hardware one20:24
Estel_transparent to OS20:24
kerio...do you think that the export gives direct access to the hardware?20:24
keriowhat the hell are you smoking20:24
kerioeverything goes through linux's IO scheduler20:24
Estel_I have no idea what would happen if two unaware things would like to write, and I don't wanna check it on my device20:24
Estel_well20:24
Estel_hope so20:24
Estel_last time I asked why the fuck maemo unmount partitions when mass storage'ing, if it work without unmounting, some people warned me about serious risk ;)20:25
Estel_I know g_file_storage is just a kernel module...20:25
kerio...because the data *is* accessed more than once at the same time20:25
Estel_well, lets imagine tracker traverse20:26
keriowhich reads20:26
Estel_and briefly checks image to filter it out20:26
Estel_hm20:26
Estel_in theory you,re right, but frankly, for what fuck I need write for kon-boot or livecd image?20:26
keriosetting it read-only is good /because you probably want it to be read-only/, not to "avoid catastrophic writes from separate things"20:26
Estel_livecd in quotes20:26
Estel_live whatever20:26
Estel_it's tailored to be run from opticaml media, which is ro too20:27
keriohell, it should be in /opt/kon-boot/something and it shouldn't be writeable by user anyway20:27
Estel_goddie idea, but are you sure that g_file_storage bootable=1 works as advertised?20:27
kerionah, there's no bootable in that20:27
Estel_maybe it would spare messing with grub20:27
keriojust make sure that the image has a MBR with a bootable flag20:28
keriomost BIOSes are picky20:28
Estel_ah, so that si why I filtered out this possibility some time ago20:28
Estel_hm20:28
Estel_sounds nice20:28
Estel_will try it asap20:28
Estel_and put you into credits (no irony) for that method if works20:28
kerio"that method" = "the way to make things boot on pcs since the 70s"20:29
Estel_you will end up in endgadget, happy? :P20:29
Estel_this method - loopback file via g_file_storage20:29
Estel_don't be silly20:29
Estel_I know about mbr ;)20:29
kerioit's just an emulated usb mass storage device :\20:29
*** RST38h has quit IRC20:29
Estel_had no idea it accept files directly20:29
keriothere's nothing else, in unix20:30
Estel_what do you want, I can't know anything before actually getting to know it20:30
Estel_so why we use swap partitions, not files :P20:30
Estel_heard they don't work for maemo swap20:30
Estel_at all20:30
kerioperhaps because of nokia's swap algorithm20:31
keriowhich still works on normal files, but makes it worse and not better20:31
keriobecause files are inherently fragmented20:31
keriofiles in a filesystem i mean20:31
Estel_shit, link for thumbified homeworld expired20:31
Estel_and gdizzz is MIA for a while20:31
kerioalso, maybe 2.6.28.10 is a bit too old to not notice the difference20:31
Estel_probably performance would hurt too20:32
keriohm, actually no20:32
Estel_but I've heard files don't work for maemo swap at all (tm)20:32
kerioin 2.6, if the swapfile isn't fragmented, it's effectively /the same thing/20:32
Estel_well, loopback image files for easy debian hurt in performance20:32
keriofor extfs, at least20:32
Estel_compared to native partition20:33
kerioif you mkfs and then make the swapfile, it's the same thing20:33
kerioloopback mount is a bit more complicated maybe? i dunno20:33
kerioswap is pretty optimized in the kernel20:33
keriohaving a partition guarantees unfragmentedness :)20:33
Estel_well, try it on your device and share results20:34
keriowhich is the big deal here20:34
Estel_sure, unless swap data written exceeds total swap size20:34
Estel_then freakin alghoritm kicks in20:34
Estel_and bye bye to consecutive linear write20:34
keriothis comes before that20:34
kerioimagine running nokia's swap algorithm on a fragmented file20:35
kerioeven the first pass will be shit20:35
Estel_but dedicated swap partition doesn't have this problem20:35
Estel_until you already written 2GB data to a 2GB file (counting one that have been freed meantime, doesn't matter)20:36
Estel_s/file/partition/20:36
infobotEstel_ meant: until you already written 2GB data to a 2GB partition (counting one that have been freed meantime, doesn't matter)20:36
Estel_unless that, it just write, ignoring free blocks in pats, where it have written already20:37
Estel_thus ~ereswap20:37
Estel_~ereswap20:37
Estel_eh, shit20:37
kerioi'm aware of what ereswap is20:37
kerioi just toggled between two partitions with a script i made for a while20:37
Estel_I'm honored to no end ;)20:37
keriothen i just merged the partitions because i couldn't be bothered20:38
Estel_I could, as I see hitting swap limit instantly20:38
Estel_by slowiness of whole ui20:38
Estel_seems to be popular experience20:38
Estel_9/10 times I'm right that I've just hit limit20:38
*** xes_ has joined #maemo-ssu20:38
kerioi seldom hit the 2gb limit20:39
kerioand usually it's the cue to make a BM backup because it's been a week or so20:39
Estel_infobot, ereswap is anti swap-fragmentation tool for maemo: http://talk.maemo.org/showthread.php?t=8441320:39
infobotEstel_: okay20:39
Estel_I hit 2GB daily ;)20:40
Estel_btw20:41
Estel_kerio, when we are at pali's bme replacement (as we are always on it)20:41
Estel_I have a device with borked usb phy or whatsnot, where charging worth ok via usb port20:42
Estel_except it need putting bq2415 into dedicated mode explictly20:42
Estel_via sysfs20:42
Estel_normally it's not a problem20:42
Estel_unless battery runs out completely20:42
Estel_as then I can't charge it unless from backupmenu (which i modified to use pali's module)20:43
keriowhy dedicated?20:43
Estel_because it doesn't detect anything from data lines20:43
Estel_so doesn't know that need to start charging20:43
keriooic20:43
kerioso... put it in dedicated mode?20:43
kerioi'd do that in /etc/event.d/bme20:43
Estel_sure, and if battery runs out, it won't charge when plugged charger20:44
Estel_thats what I was asking about20:44
Estel_where to put it to make it start charging as early as possible during boot20:44
Estel_at least before bme shuts device down due to low voltage20:44
kerio/etc/event.d/bme is the thing that loads the modules20:44
Estel_prefferably, even earlier20:44
Estel_nice20:44
Estel_huge thanks20:44
kerioyou can't quite get earlier than that i think20:44
DocScrutinizer05there's always preinit20:45
Estel_for me it isn't problem to get into backupmenu and charge a little, but device is used by my 4 years old son20:45
kerioDocScrutinizer05: good point, stuff like charging should be loaded then methinks20:45
Estel_and I won't advise him to enter backupmenu root terminal and do echo dedicated > /sys/class/(...), at least until he learns how to write, lol20:46
keriobut you still have to wait until something mounts /sys to edit sysfs nodes20:46
Estel_DocScrutinizer05,  will it work from as early as preinit?20:46
kerio...unless you can pass it as a module parameter20:46
Estel_hm, fbcon works...20:46
Estel_can't20:46
Estel_sysfs node only, sadly20:46
DocScrutinizer05kerio: the module *should* take module parameters that override defaults, on modprobing it20:46
Estel_thats why I was concerned about preinit20:46
kerioEstel_: /etc/event.d/bme is early enough for stock-ish configs20:47
DocScrutinizer05kerio: but yeah, it probably won't feel too happy without sysfs20:47
kerionah, the module will be fine20:47
keriobut you won't20:47
Estel_it's using bme replacement, thumb and things (so no stockich) shouldn't matter, though20:47
Estel_DocScrutinizer05,  btw, this device is funny re usb - twl4030 or how it's called report ok20:47
*** Vlad_on_the_road has quit IRC20:48
DocScrutinizer05kerio: actually I'm clueless how modules loaded really early handle sysfs, if there are any such modules that get loaded before sysfs mount but still do offer some sysnodes20:48
Estel_and *sometimes* dmesg report, when charger plugged, switching from state 0 to 2 and 3 and 2 and 3 again20:48
Estel_but as said port is ok and lines to testpoints are perfectly stable, low resistance20:48
kerioDocScrutinizer05: try umounting /sys and then mounting it again20:48
keriothe things are still there20:48
Estel_testpods under battery < - > usb port20:48
DocScrutinizer05kerio: brilliant :-D20:48
kerioi think20:49
keriotest it on your box!20:49
kerionothing can go wrong, surely!20:49
DocScrutinizer05i doubt sysfs can get umounted ;-)20:49
Estel_so if I would want super early, find where preinit mount /sys/ and modprobe my thing afterward?20:50
DocScrutinizer05too many open files "error: fs in use"20:50
Estel_then echo proper sysfs thing20:50
Estel_DocScrutinizer05,  use20:50
Estel_umount -l /sys20:50
Estel_:P20:50
kerioEstel_: you said that the charging doesn't work anyway20:50
Estel_lazy deattach will handle everything (tm)20:50
keriois a behaviour the same as stock ok?20:50
Estel_kerio, usb doesnt work anyway, charging work when forced20:51
Estel_yea, the same on stock or whatsnot20:51
Estel_not reflashable or coldflashable too20:51
kerioEstel_: /etc/event.d/bme20:51
Estel_upon plugging to linuxbox usb, dmesg contains nothing (on linuxbox)20:51
Estel_no change20:51
Estel_0 signs of life20:51
Estel_charge from pc usb port as dumb device20:52
*** andre__ has quit IRC20:52
Estel_otoh hardware on N900 reports ok20:52
Estel_no idea what is screwed but for 100% hardware'ishly20:52
DocScrutinizer05kerio: what are the symlinks into runlevel dirs?20:57
kerionfc20:57
kerioprobably old crap20:57
DocScrutinizer05could you check?20:57
kerioi don't know how :s20:57
keriobut afaik there's nothing for backwards-compatibility with sysv init20:58
kerioas there usually is with these fucking non-init inits20:58
DocScrutinizer05ls -l /etc/rc2.d/*boot*20:59
DocScrutinizer05ooh, that's already upstartified?21:00
DocScrutinizer05yeah, you told it, I'm dull today, sorry21:00
kerioa lot of upstarties actually call /etc/init.d/foo so i assume there's nothing generic that does that21:00
keriosee for instance /etc/event.d/rcS21:01
DocScrutinizer05I see "start on started dsme" in original events.d/bme21:01
DocScrutinizer05if the new bme is not depending on dsme, then you probably should start on <something-earlier-in-boot>21:02
DocScrutinizer05take care to split between bme-the-module and bme-the-hald-addon21:03
kerioDocScrutinizer05: meh, just setting the sysnode in /etc/event.d/bme will be enough21:03
kerioi mean, it's enough to charge stock fremantle21:03
DocScrutinizer05yeah sure21:03
DocScrutinizer05err wut? stock?21:03
kerio/etc/event.d/bme is the thing that starts bme21:04
DocScrutinizer05yes21:04
keriono charging happens before that, does it21:04
DocScrutinizer05for stock that's correct21:04
kerioit's the same thing with the bme replacement21:04
kerioand it works fine21:04
DocScrutinizer05no21:04
DocScrutinizer05since bme replacement doesn't depend on dsme21:05
DocScrutinizer05for example21:05
kerioif ! /usr/sbin/waitfordsme ; then21:05
kerioapparently it does21:05
kerionot sure for what reason, but it does :)21:05
DocScrutinizer05the module itself can get loaded/modprobed as early as sysfs exists, the latest21:05
DocScrutinizer05for what reason? cargo cult coding21:06
DocScrutinizer05as already explained above, for bme replacement we should separate bme-the-module and bme-the-hald-addon_et_al into two separate upstart jobs21:07
DocScrutinizer05bme-the-module "start on started sysfs"21:07
DocScrutinizer05bme-the-rest "start on started hal"21:08
DocScrutinizer05I do't think anything in bme replacement gets handled/started by dsme, so why would the event-job depend on started dsme?21:09
DocScrutinizer05the module doesn't depend on anything but kernel and maybe sysfs21:10
DocScrutinizer05so create a upstart job with correct dependencies21:10
DocScrutinizer05for the whole remaining bunch of cruft (hald-addon-bme, systray-battery-applet, whatnot) I dunno who starts it and what it depends on, but for obvious reasons this will need to happen later on during init. And it's not needed by bme kernel module to start charging21:12
*** sunny_s has quit IRC21:55
*** sunny_s has joined #maemo-ssu21:58
*** sunny_s has quit IRC22:11
*** sunny_s has joined #maemo-ssu22:14
*** gregoa_ is now known as gregoa22:26
*** sunny_s has quit IRC22:27
*** sunny_s has joined #maemo-ssu22:29
*** LauRoman has quit IRC22:31
*** lizardo has quit IRC22:52
*** xes_ is now known as xes22:59
*** _rd has joined #maemo-ssu23:06
*** BCMM has quit IRC23:14
*** zeq has joined #maemo-ssu23:22
*** sunny_s has quit IRC23:27
*** dhbiker has quit IRC23:28
*** sunny_s has joined #maemo-ssu23:30
*** Pali has quit IRC23:46
*** futpib has quit IRC23:51
*** BCMM has joined #maemo-ssu23:53

Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!