IRC log of #maemo-ssu for Monday, 2012-12-31

« Sunday, 2012-12-30 Index Tuesday, 2013-01-01 »
*** joshgillies has quit IRC00:27
*** nox- has joined #maemo-ssu00:37
*** dafox has quit IRC00:40
*** dafox has joined #maemo-ssu00:57
*** dafox is now known as Guest9568800:58
*** MrPingu has quit IRC01:26
*** jonwil has joined #maemo-ssu01:30
*** LaoLang_cool has joined #maemo-ssu02:04
*** LaoLang_cool has quit IRC02:07
*** NIN101 has quit IRC02:22
*** arcean has quit IRC02:24
*** nox- has quit IRC02:27
*** kolp has quit IRC02:33
jonwil~seen pali02:52
infobotpali <~pali@Maemo/community/contributor/Pali> was last seen on IRC in channel #maemo-ssu, 2d 13h 32m 26s ago, saying: 'freemangordon, do you know state of portrait keyboard?'.02:52
DocScrutinizer05jonwil: maybe you should just post your comments to the general public, pali will likely read them in backscroll/chanlog. Others might be interested and able to help as well02:53
jonwilwell its not really comments but some specific questions about changes he made :)02:55
*** joshgillies has joined #maemo-ssu03:02
*** BCMM has quit IRC03:09
*** jonwil has quit IRC03:25
*** jonwil has joined #maemo-ssu03:25
*** M4rtinK has quit IRC05:13
*** Guest95688 has quit IRC05:36
*** amiconn_ has joined #maemo-ssu05:52
*** amiconn has quit IRC05:52
*** amiconn_ is now known as amiconn05:52
*** DocScrutinizer05 has quit IRC06:03
*** DocScrutinizer05 has joined #maemo-ssu06:03
*** jonwil_ has joined #maemo-ssu07:06
*** jonwil has quit IRC07:07
*** jonwil_ is now known as jonwil07:07
*** joshgillies has quit IRC07:22
*** luf has joined #maemo-ssu09:04
lufmerlin1991: ping09:11
*** LaoLang_cool has joined #maemo-ssu09:45
*** dhbiker has joined #maemo-ssu09:50
*** LaoLang_cool has quit IRC09:56
*** dhbiker has quit IRC10:04
*** kolp has joined #maemo-ssu10:31
*** Jade is now known as jaded10:34
Sicelohmm10:40
*** dhbiker has joined #maemo-ssu11:03
*** M4rtinK has joined #maemo-ssu12:27
*** FIQ|n900 has joined #maemo-ssu13:05
FIQ|n900~changelog13:06
infoboti heard changelog is dpkg-deb --fsys-tarfile whatever.deb | tar xOf - '*changelog.Debian.gz' | gzip -d | less13:06
freemangordonluf: don't you sleep man? :P13:07
freemangordonkerio: going to try to thumb-compile libjpeg, any idea for benchmarks?13:25
freemangordonWTF?!?13:26
freemangordonno source code ?!?13:26
jon_yfreemangordon: libjpeg?13:27
freemangordonyep13:27
jon_yyou could probably find it as libjpeg6 or jpeg7 in the wild13:27
freemangordonjon_y: so? how I am sure the ABI is the same :(13:28
jon_ygood point, test it and see if an app crashes :)13:29
jon_yor look at the libjpeg headers for clues if there are any13:29
jon_yor compare symbols in the lib13:29
freemangordonsure, but this is huge amount of work13:29
FIQ|n900i wonder what HAM decides is good in terms of battery for maemo updates?13:29
*** NIN101 has joined #maemo-ssu13:29
FIQ|n900say I have a *really* slow connection13:29
FIQ|n900will it adjust?13:30
FIQ|n900or just try to install as battery runs dry13:30
jon_yI don't think it checks13:30
FIQ|n900it does check battery before doing a maemo update13:31
freemangordon:nod:13:31
FIQ|n900but *when*? after or before download?13:31
freemangordonbefore13:31
freemangordoniirc13:31
FIQ|n900genius13:31
FIQ|n900or not13:31
FIQ|n900well13:31
FIQ|n900it should check both13:31
freemangordonactually not sure13:31
freemangordonit might be after the download13:31
FIQ|n900you can stop the update when it downloads13:32
FIQ|n900but I don't think you can do other stuff when it do13:32
FIQ|n900(as checking the battery)13:32
*** povbot has joined #maemo-ssu13:40
*** ChanServ sets mode: +v povbot13:40
*** FIQ|n900 has quit IRC13:46
freemangordonperfect, libjpeg has tons of NEON optimizations made by Nokia, but no source code :(13:46
jon_ywhat is the license?13:47
jon_yif BSD/Apache, there isn't anything you can do about it13:48
freemangordonjon_y: it is even worse13:48
jon_y ?13:49
freemangordonthe license13:49
jon_ywhat is it licensed under?13:50
freemangordonit's own license, similar to BSD13:50
jon_yyeah, weak copyleft suck13:51
freemangordonI found the original code, but NEON patches are missing, ofc13:51
jon_ynot sure if asking nicely for the code will net you anything13:52
freemangordonfrom deian changelog: "-- Gary Birkett <gary.birkett@collabora.co.uk>  Wed, 10 Jan 2010 14:10:00 +0000"13:52
freemangordon*debian13:52
freemangordonI'd rather not ask ;)13:52
jon_ythe neon code came from debian maintainers?13:53
freemangordonthat guy is dead afaik13:53
*** tg has quit IRC13:53
jon_yoh, I did not know that13:54
jon_yhttp://forum.meego.com/showthread.php?t=452413:54
*** tg has joined #maemo-ssu13:54
freemangordonexactly13:55
freemangordongonna check what is the situation in meego13:55
freemangordonhmm, the same guy that changed maemo patches is the committer to libjpeg-turbo neon optimizations13:57
freemangordonwell, seems like patches are there. at least some of them:14:00
freemangordonhttp://git.linaro.org/gitweb?p=people/tomgall/meego/libjpeg-turbo.git;a=summary14:00
jon_yyou can tell what parts are missing?14:01
freemangordonhard to say if anything is missing, I have only patch names from debian/changelog14:02
freemangordonand those patches are for libjpeg-turbo, not for libjpeg14:03
jon_yare they API/ABI similar?14:04
freemangordonI may ask merlin1991 to politely ask Siarhei Siamashka for the original patches14:04
freemangordonjon_y: NFC14:04
DocScrutinizer05freemangordon: try to find other collabora (ex-)employess14:14
DocScrutinizer05~lcuk14:15
infobotwell, lcuk is a mighty good developer, best known for his work on liqbase and helping Nokia with the Identity project during the onedotzero event in London.14:15
freemangordonDocScrutinizer05: seems we need to pester that guy ^^^. And there is some :14:15
freemangordon+ * Copyright (C) 2010 Nokia Corporation and/or its subsidiary(-ies). All rights reserved.14:15
freemangordon+ * Contact: Alexander Bokovoy <alexander.bokovoy@nokia.com>14:15
DocScrutinizer05lcuk is dead14:15
freemangordon(that guy is Siarhei Siamashka)14:15
DocScrutinizer05aah14:15
freemangordonhe is the same in debian/changelog14:15
freemangordonfor maemo libjpeg neon patches, seems lcuk was just a maintainer14:16
freemangordonmerlin1991: ^^^14:16
freemangordonI am aware lcuk is dead14:17
*** BCMM has joined #maemo-ssu14:35
jonwilI always wondered why Nokia didn't open source libjpeg14:54
jonwilbut it seems the answer is "because they made nice optimizations to it they didn't want everyone else to have" :)14:55
freemangordonyeah :(14:55
jonwilSame probably applies to other stuff in SDK non-free like vorbis14:56
*** FIQ|n900 has joined #maemo-ssu14:58
FIQ|n900is >50% really "too little" for HAM?14:59
freemangordonkerio: a new thumby bb-power in in the repo15:01
freemangordon*is in15:01
*** NIN101 has quit IRC15:37
luffreemangordon: never sleeping man asking me when I sleep? :D15:59
luffreemangordon: what about bluez? Is it ok or it doesn't work for you?16:01
freemangordonhehe16:05
freemangordonno problems so far on my devel device16:05
freemangordonbut i didn;t use it much16:05
luffreemangordon: is this changelog for libpng ok?16:05
luf  * Add ARM NEON optimization patch16:05
luf    - https://codereview.chromium.org/5141003/patch/1/216:05
luf      doesn't help (even slowdown) => not used16:05
luf    - backported neon + 16-bit align from libpng15 upstream16:05
lufOr should I give their your credits?16:06
freemangordonluf: pull the latest from gitorious16:06
freemangordonI changed it16:06
lufRight I tried it last in the morning :)16:06
freemangordondebian/changelog that is16:06
lufOk.16:07
freemangordonluf: BTW if you are going to edit it, don;t increase the vesrsion number please16:08
freemangordon*version16:08
lufNo I'm not going to change it.16:08
freemangordonok. as I already pushed it i -thumb repo16:08
lufAs I wrote I checked it in the morning and there were no such change.16:08
freemangordonI know, I did it an hour ago16:08
freemangordonluf: so, zlib NEON patch brings no benefit?16:11
freemangordonBTW what path is optimized - inflation or deflation?16:12
luffreemangordon: it brings some benefit. But it's not so big as I hoped.16:12
freemangordonwell, every bit helps16:13
lufdeflation - for sure. I'm not sure about inflation.16:13
lufI'm too bored with all the benchmarks :)16:13
freemangordonBTW it might do better with gcc4.7.2, there are some comments in the patch re gcc not doing its best in some loop16:13
freemangordonluf: how did you benchmark it?16:14
lufI think also the package is -Os (maybe -O<something else> should help too.16:14
lufzpipe16:14
freemangordonWTF is that?16:14
freemangordonyeah, for ARM -02 is the pest16:15
freemangordon*best16:15
lufhttp://pastebin.com/1QWkuSS716:15
lufifneq (,$(findstring noopt,$(DEB_BUILD_OPTIONS)))16:16
lufCFLAGS=-O016:16
lufelse16:16
luf# By default, optimise for osso16:16
lufCFLAGS=-Os16:16
lufendif16:16
lufThis is from rules ... (from last maemo package).16:16
luf... and also in the package I compiled.16:16
freemangordonluf: -Os is optimize for speed16:17
freemangordonluf: where to find zpipe?16:17
*** FIQ|n900 has quit IRC16:18
lufon zlib.net16:18
lufhttp://www.zlib.net/zpipe.c16:18
freemangordonok16:18
freemangordonanyway, i'll leave that for the next year :D16:19
lufSo long time with not supporting maemo? :D16:24
lufKeep your time ;)16:24
freemangordon:D16:25
freemangordonI'm feeling lazy now16:25
freemangordonand TBH it is tklock I have to do first16:25
lufI'm glad that someone cooperate with me.16:25
freemangordonon lazyness?16:25
freemangordon:P16:26
lufYeah the question of my reboots ...16:26
lufAny hint to see what was happen?16:26
lufNow the dev N900 keeps running.16:26
lufBTW I want to upload zlib (and also recompiled libpng - againist new zlib) into cssu-devel. But I'm not sure if merlin1991 is ok with that. Also I need the patched libxml2 for it ...16:28
ShadowJKi thought -Os was size16:29
freemangordonShadowJK: could be16:30
freemangordonluf: cssu-devel is exactly for that16:30
freemangordonthough i'd expect the code to bo on gitorious before .deb lands in cssu-devel16:31
luffreemangordon: not if you and merlin1991 and others don't agree with upgrading zlib ;)16:31
lufAnd of course git repo (after agreement).16:31
freemangordonluf: anyone is against upgrading?16:32
lufyou were ...16:32
lufdue to libxml2 bug with newer zlib.16:32
freemangordonhmm, nooo16:32
lufNo way to check all the apps depending on zlib.16:32
freemangordonI was not against, I just wanted to know the possible implications16:33
luf:)16:33
freemangordonluf: i'd say put it in -devel16:34
freemangordonluf: keep in mind you have to put libxml there too16:35
*** jonwil has quit IRC16:37
lufI'll wait for merlin1991 ;) with agreement and also git repo creation.16:48
freemangordon:nod:16:48
freemangordonluf: re your reboot16:50
freemangordoninstall sp-rich-core and upload coredumps16:50
freemangordonor debug them yourself16:50
keriowhat does the sp- prefix mean?16:50
freemangordonNFC16:50
kerioit's the same as sp-oops-extract, right?16:51
freemangordonsame?16:51
freemangordonluf: http://maemo.org/development/tools/doc/chinook/sp-rich-core/16:52
freemangordonnot sure if there is an entry for fremantle16:52
freemangordonon wiki that is16:52
*** arcean has joined #maemo-ssu16:55
freemangordonluf: we should upgrade libpng in microb-engine too17:00
freemangordonor tell microb to use system libpng17:00
freemangordonhmm, it seems it already uses that]17:01
freemangordonhmm, no17:03
*** FIQ|n900 has joined #maemo-ssu17:05
FIQ|n900where can I report bugs in HAM in the read only state? not at all?17:06
freemangordonon TMO17:08
FIQ|n900hmmmmmmmm17:08
FIQ|n900in the cssu thread I suppose17:09
FIQ|n900i.e. http://talk.maemo.org/showthread.php?t=8052517:12
freemangordonyep, if you are on -testing17:12
FIQ|n900well it likely affects all cssu flavours and pre-cssu17:13
FIQ|n900but I can't test on anything else than testing17:13
freemangordonthis thread should be ok then17:13
FIQ|n900also, @ first post, latest is 7, not 5 :P17:15
luffreemangordon: what cores are you talking about?17:16
freemangordonluf: coredumps17:16
freemangordonluf: BTW we may want this http://sourceforge.net/projects/libpng-apng/files/libpng12/1.2.49/ included in libpng17:17
freemangordonso we can build microb with system libpng and zlib17:17
lufwhat coredumps? I don't see such things. The device crashed and rebooted.17:18
freemangordonluf: you need sp-rich-core installed17:18
freemangordonread the wiki article17:18
FIQ|n900posted the issue17:18
freemangordonthanks17:19
luffreemangordon: what wiki article?17:20
freemangordon http://maemo.org/development/tools/doc/chinook/sp-rich-core/17:20
FIQ|n900freemangordon: should I report apt issues in the same thread?17:21
*** Martix has quit IRC17:21
freemangordonyep17:21
FIQ|n900ok17:22
lufUgh I saw no wiki in the url. I see it and it's for next crash ...17:22
luffreemangordon: why we need apng? Any feature request?17:23
freemangordonluf: microb needs it17:23
freemangordonnow microb uses it own libpng17:23
freemangordonif we add apng support in libpng, we can build microb to use system libpng, which is NEON-optimized ;)17:24
freemangordon(will be)17:24
lufOk. However I vote for step by step ... Test the current libpng+neon patches first.17:24
freemangordon:)17:25
luffreemangordon: I'm not sure if something segfault ... I think I saw some suggestion about crash info somewhere ....17:25
freemangordonluf: install sp-rich-core and make sure you have the correct directories created. as described on wiki17:26
freemangordonif you have kenel oops, check in /dev/mtd217:26
luffreemangordon: how to check?17:26
freemangordoncat it17:27
luf/media/mmc1 is on rootfs?17:27
freemangordonno, it is uSD card17:28
lufI see no such thing in mount ...17:28
freemangordondo you have uSD17:28
freemangordon?17:28
lufNo if it's additional card.17:28
freemangordonadditional card?17:29
freemangordonmicro SD17:29
lufShould it be next to camera lens?17:30
*** FIQ|n900 has quit IRC17:30
lufI don't put any uSD. I have N900 as I bought it.17:30
freemangordonif you don't have micro SD card in the device, you should create the directory somewhere in /home/user/17:30
freemangordonnot sure exactly ehere though17:31
lufNo problem I'll symlink it to /media/mmc1/code-dumps17:31
*** FIQ|n900 has joined #maemo-ssu17:31
freemangordonluf: don;t17:31
freemangordoncreate it in /home/user/MyDocs17:31
FIQ|n900"not enough battery" with battery over 50%... ok17:32
lufmkdir /home/user/MyDocs/core-dumps/17:32
freemangordonand run some application that crash (segfaults for example)17:32
freemangordonyep17:32
FIQ|n900not my fault if apt takes an hour (literally) to load all packages subject to update lol17:32
freemangordonluf: make sure you have as many debug packages installed as possible17:32
FIQ|n900s/apt/HAM/17:33
infobotFIQ|n900 meant: not my fault if HAM takes an hour (literally) to load all packages subject to update lol17:33
luffreemangordon: It's like space shuttle launch :D17:33
freemangordonluf: only the first time :P17:33
lufI have no apps which segfault :D17:33
freemangordoncompile one :P17:33
FIQ|n900"let's break stuff intentionally" :317:34
lufUgh and you said you're the lazy one? :D I see you don't know me.17:34
freemangordon(int*)NULL = 'lif';17:34
freemangordonoops17:34
freemangordon(int*)NULL = 'luf';17:34
freemangordon*(int*)NULL = 'luf';17:34
FIQ|n900"oops, made something which actually works"17:35
freemangordonhehe17:35
FIQ|n900wow17:35
FIQ|n900disabled devel17:35
FIQ|n900and apt took 5min instead of 1h to load updates17:35
FIQ|n900though i'll still have to wait for other battery to charge17:36
*** FIQ|n900 has quit IRC17:37
*** FIQ|n900 has joined #maemo-ssu17:37
lufstrange the example from wiki doesn't crash :D17:39
freemangordonwhaat?17:40
lufOk I get it.17:40
lufAnd I see .lzo in core-dumps in MyDocs ...17:40
freemangordongreat17:40
freemangordonthat's it17:41
lufI used -O2 to be faster :D17:41
freemangordonnow you need sp-richcore-extract :P17:41
lufHmmm I have to leave now.17:41
freemangordonhttp://maemo.org/development/documentation/man_pages/rich-core-extract/17:41
freemangordonok17:41
lufSo happy new year ;) And of course a lot of enthusiasm for maemo ...17:41
freemangordonthe same for you17:42
freemangordon:P17:42
*** luf has quit IRC17:44
*** FredrIQ|n900 has joined #maemo-ssu18:09
*** FIQ|n900 has quit IRC18:11
*** FredrIQ|n900 has quit IRC18:41
*** AndrewX192 has quit IRC18:43
*** AndrewX192 has joined #maemo-ssu18:44
*** AndrewX192 has joined #maemo-ssu18:44
DocScrutinizer05kerio: sp-rich-core and sp-oops-extract are completely different tools19:05
kerioi figure19:06
*** FIQ|n900 has joined #maemo-ssu19:10
DocScrutinizer05((<freemangordon> and run some application that crash (segfaults for example))) killall -SIGBUS app19:10
FIQ|n900ah, seems HAM check both before and after download19:11
FIQ|n900anyway19:11
FIQ|n900does HAM and battery indicator use different ways of determing charge?19:11
FIQ|n900indicator showed 70%ish before download, HAM accepted and started dl19:11
FIQ|n900then refused to update once downloaded, battery still shows 70%ish19:12
DocScrutinizer05HAM may use a simplified methos not depending on HAL19:12
DocScrutinizer05like probing raw bat voltage for example19:13
DocScrutinizer05at least that's what NOLO does for deciding to allow flashing19:13
DocScrutinizer05afaik19:13
FIQ|n900ok19:13
FIQ|n900have HAM deleted downloaded file now? -.-19:15
FIQ|n900or will it actually use what it downloaded earlier19:15
FIQ|n900assuming a new update isn't out the next time I try (a hour or so from now)19:16
DocScrutinizer05I doubt it's that smart19:16
FIQ|n900:(19:16
kerioDocScrutinizer05: nah, it probably just asks hald19:17
FIQ|n900then trying to update up here is just a waste of time19:17
keriootherwise it wouldn't work if you had pali's replacement19:17
kerioDocScrutinizer05: otoh, apt caches do exist19:17
FIQ|n900and then I should wait before proper connection again19:17
kerioFIQ|n900: check /home/.apt-archive-cache19:17
DocScrutinizer05also check HAM settings in redpill mode19:18
FIQ|n900assuming you mean /home/opt?19:18
FIQ|n900DocScrutinizer05: which setting exactly?19:18
kerioassuming is bad19:18
DocScrutinizer05I seem to recall a "delete cache" option19:18
kerioit makes an ass out of u and some poor guy named ming19:18
kerioi meant /home/.apt-archive-cache19:18
DocScrutinizer05kerio19:18
DocScrutinizer05nope19:18
FIQ|n900nothing there19:18
DocScrutinizer05you meant ~user/.apt-archive-cache I guess19:19
FIQ|n900and placing it in /home sounds weird in general19:19
kerioDocScrutinizer05: just check it, dawg19:19
FIQ|n900also I don't have any of these directories19:19
FIQ|n900er19:19
FIQ|n900*also I don't have this in /home, ~ or /opt19:20
kerio\_o_/19:20
DocScrutinizer05IroN900:~# locate .apt-archive-cache19:20
DocScrutinizer05locate: Warnung: Die Datenbank "/var/cache/locate/locatedb" ist älter als 8 Tage (Das Alter ist 99,0 Tage).19:20
DocScrutinizer05/home/user/MyDocs/.apt-archive-cache19:20
DocScrutinizer05/home/user/MyDocs/.apt-archive-cache/lock19:20
DocScrutinizer05/home/user/MyDocs/.apt-archive-cache/partial19:20
DocScrutinizer05/rescue/.apt-archive-cache19:20
FIQ|n900yes I have it there19:20
kerioweird19:20
keriowhy do i have that in /home then19:20
FIQ|n900well maybe at least19:20
FIQ|n900the directory is there :p19:21
FIQ|n900and contain lots of .deb files19:21
FIQ|n900[18:18:20] <DocScrutinizer05> I seem to recall a "delete cache" option19:21
FIQ|n900"Clean apt cache" you mean?19:22
DocScrutinizer05yup19:22
FIQ|n900I have an option like that19:22
DocScrutinizer05might be related, no?19:22
FIQ|n900And it's checked19:22
FIQ|n900yeah19:22
FIQ|n900but uhh19:22
FIQ|n900it doesn't seem to work19:23
FIQ|n900as even if it's checked, I have lots of .deb files in the archive directory19:23
DocScrutinizer05so maybe it's sth different19:23
FIQ|n900but this is odd19:24
FIQ|n900latest .deb is downloaded dec 2519:24
FIQ|n900ie last week19:24
DocScrutinizer05I seem to recall there been /var/apt/cache or sth as well19:24
DocScrutinizer05once a few years ago19:25
FIQ|n900but I also have a "partial" directory, latest modified 15min ago19:25
FIQ|n900but it's empty19:25
DocScrutinizer05I once suggested to clean it, and that dude wanted to kill me for that suggestion19:26
DocScrutinizer05  /opt/var/cache/apt/archives/19:27
DocScrutinizer05 /var/cache/apt19:27
FIQ|n900DocScrutinizer05: assuming /var/cache/apt, that directory seems to exist19:27
FIQ|n900but it was modified a long time ago19:27
DocScrutinizer05 /home/opt/var/cache/apt/19:28
DocScrutinizer05well, the last one been silly19:28
kerioDocScrutinizer05: which dude wanted to kill you for that? :O19:28
FIQ|n900correction: the directory was modified today for 2h ago, but I misread19:29
FIQ|n900-but19:29
DocScrutinizer05the dude on IRC who *NEEDED* his device resurrected til next morning19:29
DocScrutinizer05but uptdate to PR1.1 failed with usual "use Nokia suite"19:29
FIQ|n900"resurrected" "update to PR1.1" wut?19:30
DocScrutinizer05I was silly enough to not clearly explain what I meant with "clean it"19:30
DocScrutinizer05FIQ|n900: loooong ago19:31
DocScrutinizer05btw get locate, it's cooolman!19:32
DocScrutinizer05and since my patch even updatedb works19:33
FIQ|n900locate being in repos?19:35
DocScrutinizer05yup20:31
DocScrutinizer05I think it's not in findutils pkg anymore, but rather own pkg. maybe not20:31
DocScrutinizer05findutils-gnu - utilities for finding files--find, xargs20:33
DocScrutinizer05locate - maintain and query an index of a directory tree20:33
DocScrutinizer05locate need findutils20:33
DocScrutinizer05I patched updatedb by adding one char "g" ;-D20:34
DocScrutinizer05so now it works with gnu-find20:34
DocScrutinizer05messybox find is brainfsckd as usual and doesn't work for updatedb20:35
FIQ|n900:P20:35
DocScrutinizer05that patch been moved to repo eventually20:35
* DocScrutinizer05 fires up a manual updatedb, just for the fun of it (I didn't feel like setting up a cron job for that)20:37
FIQ|n900Hmmm20:38
DocScrutinizer05hmmm?20:39
FIQ|n900nbm20:39
FIQ|n900*nvm20:39
DocScrutinizer05IroN900:~# time updatedb20:40
DocScrutinizer0520:40
DocScrutinizer05real    1m28.969s20:40
DocScrutinizer05user    0m49.398s20:40
DocScrutinizer05sys     0m17.164s20:40
DocScrutinizer05not half as bad as I anticipated20:40
*** FIQ|n900 has quit IRC20:43
*** freemangordon has quit IRC20:52
*** FIQ|n900 has joined #maemo-ssu20:56
FIQ|n900hm20:56
FIQ|n900something odd happened20:56
FIQ|n900instead of simply rebooting postupdate, HAM gave a two-line notice that I didn't manage to read before it rebooted20:57
FIQ|n900"The operating system has been updated. The device will reboot. Errors(...)" it said (something like it at least)20:58
FIQ|n900Did an error occur or what? Couldn't really read all of it lol20:58
DocScrutinizer05find /var ~user/MyDocs/ -mtime 121:03
FIQ|n900also once I moved some stuff from home to mydocs21:06
FIQ|n900er wrong channel21:06
DocScrutinizer05find /var ~user/MyDocs/ -mtime 1 -ls 2>/dev/null21:07
DocScrutinizer05hope you'll be able to spot something in there that has logs about that "error"21:07
DocScrutinizer05or otherwise indicates what went wrong21:08
FIQ|n900nothing21:08
DocScrutinizer05sth in syslog?21:08
DocScrutinizer05not at end but before or immediately after last boot21:09
FIQ|n900where is this?21:09
*** kolp has quit IRC21:09
DocScrutinizer05  /syslog.*restart21:10
DocScrutinizer05in less21:10
DocScrutinizer05less '+/syslog.*restart' /var7log7syslog21:11
DocScrutinizer05err21:11
DocScrutinizer05s/7///21:11
FIQ|n900no such file21:12
DocScrutinizer05hmm, then you got no syslog running21:12
FIQ|n900(and yes I corrected your spelling)21:12
DocScrutinizer05if you used / unstead 721:12
FIQ|n900yes, / instead of 721:12
FIQ|n900was what I meant :p21:13
DocScrutinizer05then you're probably out of luck to find any info about what happened21:13
FIQ|n900oh well21:13
FIQ|n900hopefully it was all ok21:13
DocScrutinizer05you could search whole device for new files21:14
DocScrutinizer05might take a while21:14
DocScrutinizer05find / -mtime 1 -ls 2>/dev/null21:15
FIQ|n900not worth the effort imo21:19
*** kolp has joined #maemo-ssu21:19
FIQ|n900because I've noticed nothing wrong21:19
*** kolp has quit IRC21:20
DocScrutinizer05sth odd with my shell foo21:20
DocScrutinizer05above line doesn't find my /var/log/syslog despite Modify: 2012-12-31 20:00:09.000000000 +010021:21
DocScrutinizer05actually above line doesn't find *anything*21:21
DocScrutinizer05:-S21:22
* DocScrutinizer05 is going to retire21:22
DocScrutinizer05nah, not really since at least on my laptop it works as expected21:22
DocScrutinizer05nfc what idiocy is built in to maemo again21:23
FIQ|n900busybox find?21:23
DocScrutinizer05IroN900:~# find --version21:25
DocScrutinizer05find (GNU findutils) 4.4.221:25
DocScrutinizer05N*F*C21:25
FIQ|n900ok21:25
DocScrutinizer05MEH!!!!21:26
DocScrutinizer05-mtime -121:26
DocScrutinizer05WTF?!21:26
*** kolp has joined #maemo-ssu21:27
DocScrutinizer05       -mtime n21:27
DocScrutinizer05              File's data was last modified n*24 hours ago.21:27
DocScrutinizer05which idiot changed sematics of -mtime in gnu-find?21:27
FIQ|n900lol21:31
DocScrutinizer05would you agree that there's not written "File's data was last modified <now>+n*24 hours" ?21:33
DocScrutinizer05ok, you cheered too soon, I won't retire. Will bash the shit out of somebody instead ;-P21:40
Raimu:)21:41
DocScrutinizer05at least if that individual changed -mtime semantics on purpose21:42
*** Estel_ has joined #maemo-ssu21:42
*** Estel_ has quit IRC21:42
*** Estel_ has joined #maemo-ssu21:42
DocScrutinizer05it took me like 3 years to wrap my mind around that somewhat weird semantics of "n*24h ago", now somebody going to change it???21:43
Estel_quick question to CSSU team, if you don't mind21:44
Estel_last version (again) changes things related to initial system boot - /usr/lib/genfstab (as before), and /etc/event.d/rcS-late, in particular21:45
DocScrutinizer05I'm afraid we'll get into same old discussion about your notion of "CSSU team" not in line with reality21:45
Estel_as person using different filesystem for /home, I had, obviously, to change ext3 to ext4 in those two files (and overall, to bring personal modifications of rcS-late to new one)21:46
ShadowJKoh hai estel, long time no see :)21:47
Estel_now, my question:21:47
Estel_hello ShadowJK :) nice to see you still here21:47
Estel_well, in fact, I was lurking here all the time, just haven't had time nor will to talk21:47
Estel_as for question...21:48
Estel_is there any other thing/config option/file/whatever, that users of different filesystems (than ext3) should consider...21:48
Estel_while using latest CSSU (thumb variant, in particular, but it shouldn't matter in this case)21:48
Estel_?21:48
* ShadowJK was considering switching to ext421:49
ShadowJKlast time I benchmarked, ext4 looked nice on flash21:49
ShadowJKnucer than ext3, that is21:49
* Estel_ is using ext4 happily, for ages21:49
DocScrutinizer05basically it was much desirable when standard CSSU files as mentioned above would already take care about non-default filesystems21:49
*** NIN101 has joined #maemo-ssu21:50
Estel_yea, benchmarks look nice. I dont't feel "obvious" improvements in real life, though - for sure no regressions, still21:50
DocScrutinizer05if they don't, it could be considered a valid bug21:51
DocScrutinizer05against CSSU21:51
ShadowJKProbably ext4 and also fixing sqlite would bring a major reduction in jitter and stutter21:51
Estel_fun fact - last time CSSU update put ext4/reiserfs into reboot loop, I was the one who reported reasons for it (found them during debugging session with freemangordon, as I was blaming CSSU). Now, months later, when installing CSSU on family member's device, I was hit ny reboot loop - just to find out, after day of searching, my own report about reasons :P21:52
Estel_ShadowJK,  define fixing sqlite?21:52
Estel_also, I've heard about some parameters via tuenefs, that we should set to make ext4 even more efficient in our device's case... (ones that work only on ext4, not older ext's), but never tried them21:53
Estel_something about fragments and blocks? /noob mode21:54
FIQ|n900what about disabling journal21:54
ShadowJKFIQ|n900; no21:54
FIQ|n900ok21:54
ShadowJKif anything, we want an absolutely massively humongous journal21:54
FIQ|n900i've heard there's no point on it in n90021:54
DocScrutinizer05btw the above question is a tad fuzzy and hard to answer, at least it's not exactly related to CSSU21:55
Estel_well, journal saved my as* dozens times21:55
DocScrutinizer05(any other files)21:55
ShadowJKWell, none of the file integrity procedures in any filesystem (besides ubifs) works on flash anyway21:55
Estel_well, I mean only files related to last CSSU update21:55
Estel_I've found two files mentioned above, that need manual tweaking from recovery shell to allow device boot...21:56
FIQ|n900..then why should you keep it?21:56
FIQ|n900@ ShadowJK21:56
DocScrutinizer05file a bug, they should allow boot from arbitrary filesystems when we already touch them21:56
Estel_still, no idea, if I should also change something else, to avoid non-critical (not putting me into bootloop) functions of latest cssu update21:56
ShadowJKSo journal or no journal from file integirty point of view probably makes little difference, although having journal present stops fsck, and fsck gets confused by flash dropping random megabytes here and there21:56
FIQ|n900ok21:57
ShadowJKBut journal is written sequentially, which gives us speed boost. Of course, later when journal is comitted we take the penalty for writing non-sequential..21:57
Estel_DocScrutinizer05,  I can, but wasnt it you who replied - last time - that CSSU won't ever support different filesystems, officially?21:57
DocScrutinizer05ShadowJK: the trick is to write-back  journal during times when system is idle21:58
DocScrutinizer05err, if anything I answered CSSU can't support user mods on system files21:58
ShadowJKAs for sqlite, same thing really, allegedly for harmattan/meego they were investigating a mode that essentially made sqlite journal-only21:59
DocScrutinizer05which might sound similar to your ears21:59
Estel_ShadowJK,  what you have written may be reason, why my home gest TOTALLY fcked sometimes, when small thing gets wrong (i.e. one file get busted wrong way by crash, and fsck gest confused = wipe out half of my inodes)?21:59
ShadowJKFor sure, sqlite writing one row/record/whatever and calling fsync 5 times takes about 5 secs21:59
Estel_(then fail to repair anything, as e2fs is ancient on N900, and segfaults during restoration phase...)22:00
DocScrutinizer05if I actually answered CSSU will never support non-default filesystems then I uttered BS22:00
ShadowJKYeah I tried fsck on /home once, big big mistake :)22:00
FIQ|n900speaking of fsck, is fscking MyDocs safe?22:00
ShadowJKif it's unmounted22:00
FIQ|n900well it isn't but i was going to umount it before doing it so.. :p22:00
DocScrutinizer05fsck of MyDocs never really worked, it takes AGES22:01
DocScrutinizer05like... DAYS22:02
Estel_well, ShadowJK, guy named BadHabit released, one, actual Debian's e2fsprogs (that don't segfault on N900) - but, still, it won't work *really* well on N900, due to flash?22:02
FIQ|n900DocScrutinizer05: hmm..22:02
Estel_DocScrutinizer05,  hm?22:02
ShadowJKEstel_; I'd say so22:02
FIQ|n900i was going to remove bogus files in mydocs22:02
FIQ|n900which it didn't allow me to22:02
Estel_I'm doing fsck.vfat /dev/mmcblk0p1 many times (MyDpocs unmounted)22:02
FIQ|n900due to "IO errors"22:02
Estel_never got any problems with it?22:02
DocScrutinizer05well, it might work occasionally, depending on amount of mess and number of files there22:03
ShadowJKFIQ|n900; check dmesg also for physical io errors :s22:03
DocScrutinizer05problem is RAM for fsck22:03
DocScrutinizer05eventually enters swap hell22:03
DocScrutinizer05then everything comes to a grinding halt22:04
DocScrutinizer05while fsck on a clean fs should usually work22:04
FIQ|n900ShadowJK: "[ 4910.734466]     invalid access to FAT (entry 0x350d7ff6)" like these?22:04
ShadowJKI think I've only done it via pc22:04
ShadowJKthose are filesystem errors22:05
DocScrutinizer05via PC should work22:05
FIQ|n900this started by a system freeze when I mv-ed some files from ~22:05
Estel_hm, I remember fsck.vfat repairing errors on my microSD vfat, once.  Never took ages. Any community project touched fsck.vfat?22:05
FIQ|n900"[ 4910.734436] FAT: Filesystem error (dev mmcblk0p1)" is shown numerous times when I ls the directory22:06
Estel_also, keep in mind that I have radically uncommon swap settings ( ShadowJK would sympatize)22:06
DocScrutinizer05as explained above it depends on swap usage and number of errors to fix22:06
* ShadowJK has swap+ext3 on uSD :-)22:06
DocScrutinizer05uSD has a hw interface independant of swap22:06
ShadowJKgpodder's .gpodder on uSD too, so that /home is idle when gpodder downloads to Mydocs :)22:07
Estel_:) 2x 2GB swap on microSD (only one used at once), and very uncommon settings re swap I/O, batching, etc22:07
FIQ|n900also now by listing the directory, MyDocs seems to have remounted to read-only22:07
FIQ|n900s/listing/removing/ (trying to)22:08
DocScrutinizer05FIQ|n900: use mass storage mode, do fsck from PC22:08
DocScrutinizer05I never did it but I heard it works, and usually flawlessly22:08
Estel_well, anyway, my question about any other things related to different (than ext3) filesystem present on device, and last cssu release, still apply. I wonder who touched upstart and ke-recv in last CSSU?22:09
Estel_ke-recv is for sure Pali's work22:09
FIQ|n900"FATs differ but appear to be intact. Use which FAT ?" what?22:09
FIQ|n900DocScrutinizer05: hmm22:09
DocScrutinizer05FIQ|n900: difficult22:09
Estel_noticed new panic settings in rcS-late, merged them with my rcS-late. obviously, updated genfstab. I wonder, if something else should be considered.22:10
ShadowJKFIQ|n900; you've got a backup of before-fsck, right..22:10
Estel_FIQ|n900,  most of the times, use older one22:10
FIQ|n900ShadowJK: tbh no, as I don't really have anything important in it22:10
FIQ|n900unless maemo settings go there, which I was unaware of if it did22:11
Estel_being scientifically curious about results?22:11
ShadowJKHeh, has that "fsck if mount fails" in rcS-late always been there, or is that cssu?22:11
DocScrutinizer05FIQ|n900: my suggestion: copy everything to PC, mkfs.vfat, copy PC->MyDocs22:11
Estel_always been there22:11
*** NIN101 has quit IRC22:11
DocScrutinizer05ShadowJK: I already blamed a massive bug about it22:12
FIQ|n900"copy everything to PC" as in, dd-ing the fs to a file?22:12
DocScrutinizer05it doesn't show ANY idication to user about fsck ongoing and possibly slowing down boot to minutes or hours22:12
DocScrutinizer05FIQ|n900: copy as in cp -a22:13
ShadowJKWell also in my experience fsck is more harmful than no fsck :s22:14
DocScrutinizer05FIQ|n900: dd would also copy the borked fs metainfo22:14
ShadowJKmaybe fsck was added after optification stuff22:14
Estel_exactly22:14
Estel_especially which parts of procedure segfaulting ;)22:14
Estel_always segfaulting, btw22:14
FIQ|n900DocScrutinizer05: that's my point, wasn't the goal to clean the fs on computer?22:15
DocScrutinizer05if fsck.ext2 is segfaulting, please rise a bug against CSSU, it needs to get fixed22:15
DocScrutinizer05FIQ|n900: nah, copy stuff to PC, mkfs on N900, copy stuff back to N900 to the virgin fs22:16
FIQ|n900got it22:16
DocScrutinizer05will fix fs as well as defragment and create a backup on PC22:16
FIQ|n900though i'll wait a while and eventually come back asking next week as I don't have computer here22:16
DocScrutinizer05and most likely be faster than fsck22:16
FIQ|n900and just try to avoid using mydocs meanwhile22:17
DocScrutinizer05well, then try fsck22:17
DocScrutinizer05on device22:17
FIQ|n900now?22:17
FIQ|n900ok22:17
DocScrutinizer05abort if it fails to finish in ~30min22:17
FIQ|n900FATs differ but appear to be intact. Use which FAT ? -- what does this mean?22:18
FIQ|n900(I rarely if ever use fsck)22:18
DocScrutinizer05there are FAT copies, if one fails22:18
FIQ|n9001:use first FAT, 2:use second22:18
Estel_well, there are certain cases, when another sub-program need to be called22:18
Estel_something about lost+found?22:18
Estel_it's part of fsck procedure in this case22:18
ShadowJKfsck would call fsck.ext322:18
Estel_so, it doesn't fail *always*, it segfault when fsck try to repair serious things22:19
DocScrutinizer05FIQ|n900: up to you which to use, probably 1st is the one which alreeady has files deleted22:19
ShadowJKbut that's about it afaik22:19
DocScrutinizer05while 2nd hasn't yet22:19
Estel_ShadowJK,  then 2nd pass comes22:19
Estel_damn how was it called22:19
Estel_gimme a sec22:19
FIQ|n900hmm wait, battery low22:19
FIQ|n900i'll screw this for the coming 20min22:20
DocScrutinizer05charger?22:20
FIQ|n900I use an external charger22:20
DocScrutinizer05or broken USB?22:20
DocScrutinizer05aah :-/22:20
*** kolp has quit IRC22:20
FIQ|n900well i have 2 batteries, so just waiting until this one dies before switching and resuming trying to fix mydocs22:21
*** NIN101 has joined #maemo-ssu22:21
DocScrutinizer05Estel_: I'm not aware of any other files that would need special care when fs-type of /home got changed22:22
DocScrutinizer05for fs-type of MyDocs you might need to check the scripts that export mass storage22:23
Estel_thanks. Yea, aware of mydocs (using fat there, anyway)22:23
DocScrutinizer05yw22:23
Estel_e2fsck22:23
DocScrutinizer05no warranty though22:23
Estel_this segfaults on 2nd pass22:23
Estel_if directories/files need 2nd pass22:24
DocScrutinizer05I'll blame werner almesberger ;-P22:24
Estel_damn, I'm quite bad at explaining it. Go, make backuo, then do a mess in filesystem and try fsck on device :P22:24
DocScrutinizer05ehmmm, maybe I already did?22:25
DocScrutinizer05(blame wpwrak)22:25
DocScrutinizer05seem to recall discussions with him about bugs in fsck22:25
DocScrutinizer05oooh, fsck.vfat22:26
DocScrutinizer05iirc22:26
Estel_I would quote exact way to replace faulty Maemo's version with debian one, but *of course* I can't scroll my PM box to second page, on device, due to awesome TMO theme, supporting 1024 as lowest22:26
Estel_and cutting half of buttons, no matter of zoom22:26
DocScrutinizer05>>dosfstools  were  written by Werner Almesberger<<22:26
DocScrutinizer05man fsck.vfat22:27
DocScrutinizer05>>This version of e2fsck was written by Theodore Ts'o<<22:27
*** FredrIQ|n900 has joined #maemo-ssu22:36
FredrIQ|n900well ok22:36
FredrIQ|n900battery died22:37
FredrIQ|n900now to fsck MyDocs22:37
*** FIQ|n900 has quit IRC22:37
*** FredrIQ|n900 is now known as FIQ|n90022:38
FIQ|n900wow MyDocs seem to be really messed up22:42
FIQ|n900after a while of asking whether to drop files with bad names, it ended up asking "Free cluster summary wrong (257837 vs. really 262121)" and prompted for correct or don't correct22:44
FIQ|n900both answers led to aborting the operation completely22:44
FIQ|n900the last output from it before this was "  Start cluster beyond limit (4294967295 > 442378). Truncating file."22:45
*** Estel_ has quit IRC22:48
FIQ|n900so DocScrutinizer05 what do you suggest?22:48
*** Estel_ has joined #maemo-ssu22:49
*** Estel_ has quit IRC22:49
*** Estel_ has joined #maemo-ssu22:49
FIQ|n900another question22:58
FIQ|n900is user settings stored in MyDocs or just media data?22:58
kerioapplication-dependent22:58
keriousually in ~/.something22:59
FIQ|n900because if it's the latter I might as well backup critical stuff (i.e. backupmenu entries for example) and just reformat22:59
FIQ|n900hmm ok22:59
FIQ|n900I tried fsck -a atm22:59
FIQ|n900because that one allowed me to proceed after that odd question which simply aborted no matter what I did23:00
keriorun backupmenu or rescueos, export the whole emmc via mass storage, fsck from a computer23:00
FIQ|n900broken usb23:00
FIQ|n900and no computer access atm anyway23:00
FIQ|n900ok after 10min fsck -a exits with error: "Unable to create unique name"23:10
DocScrutinizer05FIQ|n900: you have a uSD?23:11
FIQ|n900hmm23:13
FIQ|n900yes but not sure if inserted23:13
FIQ|n900sec23:13
FIQ|n900no23:13
FIQ|n900then I don't have it23:13
FIQ|n900(unless I wait a week)23:13
DocScrutinizer05copy everything (that still exists after that fsck run) to uSD (don't forget to take care about unfreezing measures for large cop^K23:14
FIQ|n900I don't think fsck has done anything at all..23:15
FIQ|n900it simply exited with no apparent operation23:16
FIQ|n900DocScrutinizer05: no uSD until 6th23:16
ShadowJKi think it needs -w to actually do stuff :)23:16
FIQ|n900also anti-freeze is already in place23:16
FIQ|n900uhm default is "simulate"?23:17
FIQ|n900or? :p23:17
FIQ|n900ah it does figure what to do first, then write to disk23:19
ShadowJKi seem to remember it's funny23:19
FIQ|n900but as it failed, nothing altered23:19
FIQ|n900unless -w (write immediately) is set23:20
DocScrutinizer05o/23:20
DocScrutinizer05and happy new year23:20
FIQ|n900same23:20
FIQ|n900so..23:20
FIQ|n900hm23:20
FIQ|n900i'm kinda stuck in current state atm as I'm not home23:21
FIQ|n900also interesting is that MyDocs seems mostly intact, as I discovered the corruption just now (I used to think it simply copied half of the stuff then nothing)23:26
FIQ|n900which is a month since i did this...23:27
FIQ|n900and mydocs is in rw unless I show or try to modify these files23:27
FIQ|n900so imo it *should* be possible to solve just by removing them (which seems impossible)23:28
*** nox- has joined #maemo-ssu23:32
FIQ|n900or should have been*23:36
FIQ|n900as it was a long time ago, fs becomes more messed up23:37
FIQ|n900problem appears to be solved by fsck -w23:46
FIQ|n900and by the looks of it, file system still contains the stuff I'm interested in (besides the file I mv-ed last month, but that was kind of obvious)23:47
FIQ|n900reason for my willing to continue fix even though I risk have lost stuff is that I didn't want to damage the fs even more by continuing to use it23:48
DocScrutinizer05congrats23:56
FIQ|n900ty23:58
« Sunday, 2012-12-30 Index Tuesday, 2013-01-01 »

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