IRC log of #maemo-ssu for Saturday, 2012-03-24

*** NIN101 has quit IRC00:14
*** M4rtinK has joined #maemo-ssu00:41
*** BCMM has joined #maemo-ssu00:41
*** Pali has quit IRC00:59
*** BCMM has quit IRC01:04
*** BCMM has joined #maemo-ssu01:04
*** BCMM has quit IRC01:05
*** BCMM has joined #maemo-ssu01:08
*** Jade has quit IRC01:27
*** zogg_ has joined #maemo-ssu01:44
*** nox- has quit IRC01:55
*** ZogG has quit IRC01:55
*** DocScrutinizer51 has quit IRC01:55
*** Jaffa has quit IRC01:55
*** nox- has joined #maemo-ssu01:59
*** BCMM has quit IRC02:16
merlin1991DocScrutinizer: how did it go?02:21
DocScrutinizernot at all yet, chillout failed. Somebody thought it's a good idea to get me even more upset by pestering me about infobot02:22
*** trumee_afk is now known as trumee02:26
*** trumee is now known as trumee_afk02:31
*** M4rtinK has quit IRC02:34
*** trbs has quit IRC02:40
merlin1991arcean: do you have time for another meeting next wednesday after 20 utc?02:43
*** LaoLang_cool has joined #maemo-ssu02:52
*** LaoLang_cool has quit IRC02:58
*** LaoLang_cool has joined #maemo-ssu02:59
arceanmerlin1991, ok, no problem03:04
*** ChanServ sets mode: +o merlin199103:08
*** merlin1991 changes topic to "Maemo Community Seamless Software Update "CSSU" channel, | Known bugs: | Channel logs: | Sources: | Latest version (testing): 21.2011.38-1Tmaemo2.1; (stable): 21.2011.38-1Smaemo2.1 | meeting 28.03 20:00 UTC"03:09
*** merlin1991 sets mode: -o merlin199103:09
arceanmerlin1991, thanks and bye :)03:12
*** int_ua has joined #maemo-ssu03:13
*** arcean has quit IRC03:14
*** int_ua has quit IRC03:36
*** int_ua has joined #maemo-ssu03:54
*** chainsawbike has quit IRC04:15
*** chainsawbike has joined #maemo-ssu04:21
*** amiconn_ has joined #maemo-ssu05:21
*** amiconn has quit IRC05:21
*** amiconn_ is now known as amiconn05:21
*** nox- has quit IRC05:37
*** Jade has joined #maemo-ssu05:53
*** Milhouse has quit IRC07:31
*** Milhouse has joined #maemo-ssu07:35
*** andre__ has joined #maemo-ssu10:05
*** andre__ has joined #maemo-ssu10:05
*** M4rtinK has joined #maemo-ssu10:59
*** trumee_afk is now known as trumee11:05
*** NIN101 has joined #maemo-ssu11:06
*** Kaptenen has quit IRC12:02
*** Pali has joined #maemo-ssu12:05
*** DocScrutinizer has quit IRC12:45
*** DocScrutinizer has joined #maemo-ssu12:45
*** arcean has joined #maemo-ssu13:23
*** LaoLang_cool_ has joined #maemo-ssu13:34
*** LaoLang_cool has quit IRC13:34
*** arcean_ has joined #maemo-ssu13:51
freemangordonDocScrutinizer, what about my nick?13:51
freemangordonDocScrutinizer, pong13:52
*** arcean has quit IRC13:53
psycho_oreosfreemangordon, ping13:54
psycho_oreosyou were wanting me before?13:54
freemangordonpsycho_oreos, yeah, you had some issue you were thinkong it might be casued by CSSU, how it went?13:55
psycho_oreosfreemangordon, ahh it wasn't related to CSSU, DocScrutinizer cleared that up as it was an issue with autodisconnect. I got some help from #maemo where we actually found the culprit13:56
freemangordonok, good13:57
freemangordonI'll search the logs13:57
LaoLang_cool_Is CSSU testing stable enough for daily usage?14:02
psycho_oreoswhat's wrong with the current stable version?14:03
freemangordonpsycho_oreos, it is missing lots of features from -testing14:04
freemangordonLaoLang_cool_, that is what I am using on my only phone, so I assume it stable14:04
psycho_oreosfreemangordon, heh if the user wants stable, there's stable, otherwise there's testing and that's for those who risk it just to be on bleeding edge14:05
LaoLang_cool_freemangordon: wow, good :)14:05
psycho_oreosI'm sure that sort of rule applies to anything one tries, you either stick with stable and be happy or you can run the risk for the added benefits and go with bleeding edge14:05
psycho_oreosI'm also using the latest testing as well14:06
freemangordonpsycho_oreos, actually -testing is not -devel ;)14:06
Lava_Croft-testing is kind of like the area for release candidates14:06
psycho_oreosfreemangordon, though in CSSU terms -testing == -devel :)14:06
freemangordonso there is very little chance something to be totally broken14:06
freemangordonit is not -devel14:06
psycho_oreosso there's actually a devel CSSU? :)14:07
Lava_Croftit certainly is not14:07
Lava_Croftif it was -devel, i wouldnt never use it14:07
* psycho_oreos thinks of that gitorious repo?14:07
infobotpsycho_oreos meant: so there's actually a devel CSSU\. :)14:07
LaoLang_cool_though it'd called testing, but its updating is not so active, so it can be thought as stable14:07
freemangordonpsycho_oreos, ask merlin1991 for "devel-repo-to-be"14:08
psycho_oreosfreemangordon, I'll do so when I see merlin199114:09
*** arcean_ is now known as arcean14:11
*** arcean_ has joined #maemo-ssu14:16
*** arcean has quit IRC14:18
*** arcean_ is now known as arcean14:19
*** arcean_ has joined #maemo-ssu14:21
*** arcean has quit IRC14:23
*** BCMM has joined #maemo-ssu14:37
*** BCMM has quit IRC14:44
DocScrutinizerhi folks, what about - we already got the patched libsms in CSSU?15:01
povbotBug 8347: Cell Broadcast Feature not available15:01
DocScrutinizermerlin1991: ^^ ?15:01
DocScrutinizerit's thoroughly tested and seems to work without *any* problems15:03
DocScrutinizerAIUI it's part of this new provider thingie15:05
DocScrutinizerthe jonwil thingie15:10
PaliDocScrutinizer, that patched binary is licensed under nokia closed15:10
Palishould we ignore license arguments and push patched binary to CSSU?15:10
PaliI think that Nokia does not care about that now...15:10
DocScrutinizerso we either ignore that, or deliver the patch proggie (a c 5 liner I wrote)15:11
DocScrutinizerrather I provided the template, it's been written by jonwil I guess15:11
PaliI think the best should be to push binary deb package which contains that library to CSSU (with patched binary)15:11
Paliand really ignore nokia license15:12
Lava_Croftstand up against the man15:12
DocScrutinizerseems we're already shipping other nokia binaries, no?15:12
Paliquim wrote something about that we can redistribute nokia closed binaries for open projects15:12
PaliRE binaries :D (l10n)15:13
DocScrutinizerwhich btw IIRC was one of the reasons to get CSSU on Nokia server15:13
PaliDocScrutinizer, only l10n binaries we have yet15:13
Palinothing more15:14
DocScrutinizer:nod: it was too much hassle and too little benefit for the certman stuff15:14
Palicertman is now full open :D15:14
PaliI forced nokia to open it :D15:14
Palithe only stuff which was open for fremantle15:15
PaliDocScrutinizer, what is needed to test cell broadcast?15:16
DocScrutinizeranyway that libsms has a severe and clearly located bug, with a verified fix15:16
DocScrutinizerjonwil's provider/location thingie for status area15:17
DocScrutinizeror my python proggie15:17
DocScrutinizerplus of course the patched lib15:17
DocScrutinizerthere's a thread in tmo about it15:18
DocScrutinizerdon't ask me where15:18
Palishould be?15:18
DocScrutinizererrr, dunno, the lib is called libsms or licisisms or sth like that15:19
DocScrutinizerjonwil provided all it needs to get the stuff into CSSU, and I honestly thought we included it long ago15:20
Paliok, I suggest to create same DEB packages which are on with increased version15:21
Palishould I start creating dsc?15:21
DocScrutinizermerlin1991: ^^^15:25
DocScrutinizermerlin1991: what's state of operator-name-cbs-widget?15:25
DocScrutinizerdoes it ship with patched libsms, or  patch-libsms binary?15:26
DocScrutinizerdoes it ship at all with latest T?15:27
DocScrutinizerjonwil (and I) are working on this since felt years15:27
DocScrutinizermerlin1991: I want to add a comment to "fixed in CSSU"15:28
povbotBug 8347: Cell Broadcast Feature not available15:28
DocScrutinizeractually it's been >1 year15:29
DocScrutinizerabsolutely what CSSU is meant for15:30
DocScrutinizerforgetting to include it in recent CSSU-T was rather rude against jonwil15:32
DocScrutinizerthe patch for libsms has to go into CSSU anyway, no matter what's the decision about operator-name-cbs-widget15:34
DocScrutinizersee rationale above15:34
DocScrutinizerlibsms has a severe and clearly located bug, with a verified fix15:35
* amiconn likes cssu as such, but hates its update procedure15:45
*** rlinfati has joined #maemo-ssu15:45
amiconnIt always (1) loses packages (used to be three, down to one now but still annoying and (2) disables bluetooth15:46
rlinfatiflash10 can be accepted in cssu ?15:51
rlinfatiadobe flash player15:51
NIN101it would be news to me that adobe flash 10 exists for maemo...15:55
DocScrutinizerthere's been that "pirated" stuff from TI, which proved to be worse than original flash9 incl special Nokia patches15:56
DocScrutinizeramiconn: which pkgs get lost?15:56
merlin1991DocScrutinizer: it was supposed to end up in latest testing, but mag had some build errors15:58
merlin1991so we postponed to next testing15:59
*** ruskie has quit IRC15:59
merlin1991which hopefully arrives around next friday15:59
rlinfatiand the flash9 with the version string changed to report 10 or 11 ?15:59
DocScrutinizernext friday which universae?15:59
merlin1991I know15:59
merlin1991currentlye queued for next testing is krev-c, alarmd, libxml, and the smslib + widget16:00
NIN101rlinfati: I don't think CSSU would mess with flash version strings...16:00
merlin1991we're not going to mess with flash version string ever16:00
NIN101see :-).16:01
merlin1991there is a package for that in -devel, dunno if it went up to extras16:01
DocScrutinizerI'm not that averse16:01
DocScrutinizerthere might be good reasons to patch the version string16:01
merlin1991and that would be?16:01
DocScrutinizerwebsites querying the string16:02
merlin1991in any case announcing I am version x though in reality you are y is bound to break at some point16:02
merlin1991if the user wants to take that risk there is a package for that in -devel16:02
DocScrutinizerin fact what we got is more 10 than 916:02
merlin1991not something that should be forced on the general16:02
merlin1991s/general/general user/16:02
infobotmerlin1991 meant: not something that should be forced on the general user16:02
DocScrutinizerafaik at least16:03
DocScrutinizer9 plus massive patches16:03
DocScrutinizerwhich actually are backported 10 features16:03
DocScrutinizerso it's maybe 9.7516:04
PaliDocScrutinizer, merlin1991 - I'm creating correct debian/* stuff for that cellular libconnui packages16:04
DocScrutinizeryou might argue a verstr=10 matches better16:04
Paliit will produce same package name as in nokia repository16:05
merlin1991anybody who wants to edit his version string can use
DocScrutinizerPali: I dunno about libconnui16:05
DocScrutinizermerlin1991: fair point16:05
merlin1991it's even in extras :)16:05
DocScrutinizeractually there's nothing in CSSU that would suggest we need to fix the version string, for now16:06
DocScrutinizerI'm a bit deciding this based on my gut feelings, which so far were good enough for companies to listen to me16:07
merlin1991amiconn: which package did you loose this time around?16:07
merlin1991btw DocScrutinizer did you update already?16:08
DocScrutinizerlibsms: part of system, closely woven in by API calls from other parts. flash version: not visible to system at all16:08
DocScrutinizerI sometimes have a hard time to explain why I decide one way or another, but when and after I do, it becomes clear usually that there's sound reason16:09
DocScrutinizerof course that'S not a basis for discussion when somebody challenges my decisions16:10
DocScrutinizerotoh we're on same page for now, regarding libsms as well as flash16:10
*** ruskie has joined #maemo-ssu16:11
DocScrutinizeras soon as massive reports come in that e.g. youtube refuses to work with unpatched flashlib while it works with patched flash version string, I'd prolly change my mind16:12
DocScrutinizeras then it's obvious the system as a whole has a "bug" which can get fixed easily16:12
DocScrutinizerand again I like to push for a extras-cssu repo, that has apps tested on and tailored for CSSU16:14
DocScrutinizerfor now we already have ONE app in there: orientation-lock-widget16:14
merlin1991that's not going to happen on maemo.org16:14
DocScrutinizer:shrug: I'm passionless about where that repo lives16:15
DocScrutinizerwe probably not even need a new repo, just similar logic/stics like extras, where some pkgs are not visible via HAM and still can get installed16:17
DocScrutinizerthis class would make our CSSU core16:17
DocScrutinizerwhile the visible pks are the optional ones16:17
DocScrutinizerdon't you think this can work?16:18
DocScrutinizerwe also might want to move e.g. nicocam to that optional class16:21
DocScrutinizerwhich solves our problem of "not possible to rollback"16:21
DocScrutinizeras those CSSU installations out there that already have nicocam can happily pull updates from "cssu-optional" class when next CSSU rev rolls out, while those installing fresh wouldn't automatically include nicocam any more16:23
DocScrutinizeractually AIUI our CSSU is a MP and visible in HAM from CSSU repo, while the pkgs inside that MP are not visible and also live on same CSSU repo. Right?16:27
DocScrutinizermove nicocam and orientationlock-applet out of MP and into an own class similar to CSSU-MP, et voila16:28
DocScrutinizerclass here means HAM app class, that decides about visibility and category16:28
DocScrutinizerI'm not into HAM that much to know how that's implemented, but I heard it's several classes some of which HAM actively filters and doesn't display16:29
DocScrutinizerunless you put HAM into redpill mode ;-D16:30
DocScrutinizerPali: you #maemo ping already serviced?16:31
Palimerlin1991, see this:
PaliDocScrutinizer, is there a way to get usb speed when hostmode is active?16:38
Palior it is impossible?16:38
DocScrutinizerfind out which speed?16:38
Paliautodetect low/high/full mode16:39
DocScrutinizerwell, first of all you have to actively set speed when using N900 hostmode16:39
DocScrutinizeras you can't change speed as soon as vbus got attached16:39
DocScrutinizerand you can't detect speed of periipheral without vbus16:39
Paliso, not possible?16:40
DocScrutinizerthat'S the whole reason why h-e-n has speed selector, and why autospeed detection doesn't work16:40
DocScrutinizerkernel detects speed of attached peripheral on vbus engaged, but then kernel can't switch speed accordingly anymore16:41
DocScrutinizerand since vbus isn't controlled in kernel domain, you have to implement all "auto-speed" in h-e-n or a similar userland app that enables vboost, reads out what kernel found out about peripheral, then removes vbus for long enough to terminate the USBG session, sets speed accordingly and ramps up hostmode incl vboost again16:43
DocScrutinizerblueled's script is trying to do exactly that16:43
DocScrutinizerPali: does this explanation answer your questions?16:45
DocScrutinizerI may add that this inability to switch speed with vbus attached is an idiosyncrasy of exploited testmode we had to resort to, to make hostmode possible at all on N90016:46
PaliAs vbus is controled by my bq2415x driver in kernel, I'm thing if it can be possible to autodetect speed also in kernel16:47
DocScrutinizeron a proper usual usb-hc, the speed is switched automatically even in musb-core hardware block, and kernel only takes notice but doesn't actively switch speed16:47
DocScrutinizerso even if you'd decide to try auto speed-detect in kernel, you still are far away from how USB *usually* works16:48
DocScrutinizerplus you get layering issues when trying to control vbus boost in bq24150 from musb-hdrc, I'd guess16:49
Palicommunication between bq2415x and isp* modules are done via platform data in board-rx51.c16:50
Paliso some hook can be added into musb*.c to call some platform_data function16:51
DocScrutinizerand since kernel is .. well kernel, you lose control of what it actually does regarding vboost and charging and hostmode etc, means you'll have a hard time to recover when any of this stuff doesn't behave with a particular paripheral the way you (the kernel devel who designed all this) thought16:51
DocScrutinizerI strongly suggest not to implement anything like this into kernel16:52
Palibq2415x driver is fully configurable via sysfs16:52
DocScrutinizersince kernel still ahs to cope with bme, if nothing else suggests not to do it16:52
Palibq driver does not work with BME :D and it replacing bme16:54
DocScrutinizeryou'll have an incredibly hard time to make your kernel compatible with bme, even if you'd do "the right thing" and replace the vboost-enable function already existing in kernel (but empty function, formerly controlled twl4030 chargepump) by something that directly drives bq2415016:55
merlin1991Pali: I can't merge that mg16:55
DocScrutinizerthat'S exactly why we decided NOT to do it like this when we patched kernel for h-e-n16:55
merlin1991the problem is libconnui is not part of the fremantle scratchbox, so you can't link against it without the binary in the -dev package16:56
Palimerlin1991, ah ok16:57
DocScrutinizerPali: to make a proper hostmode kernel with vboost work for a standard maemo fremantle system with bme, you'd need such massive nasty hacks to avoid bme interfere with vboost mode. It's not feasible16:57
Palibut why with BME?16:57
Palibq24* driver is replacing BME16:58
DocScrutinizerbecause bme gets started in fremantle initscripts, and is needed by e.g hal-addon-bme etc16:58
DocScrutinizerof course you can build a proper hostmode kernel, but that one will be incompatible to fremantle16:59
Palihal-addon-bme can be replaced16:59
DocScrutinizerwell, actually a *proper* hostmode kernel doesn't even work on N90016:59
Paliinterface to HAL is known and bq241* interface to kernel too16:59
DocScrutinizeryes, it can. but not for stock fremantle16:59
DocScrutinizerplus, as elaborated above, the fsckdup N900 hw doesn't allow a really *proper* hostmode anyway. You always need to exploit testmode, with all the negative side effects as described above17:01
DocScrutinizeryou need to implement into kernel what blueled's hostmode script is doing in userland right now. And you see this script doesn't work reliably. But at least you can reliably stop it17:02
DocScrutinizeryou can't stop your kernel when it starts to act up17:02
merlin1991oh you can, pull out the battery ;)17:03
DocScrutinizerhonestly, I've been there, done that (or rather, decided not to do)17:03
*** LaoLang_cool_ has quit IRC17:03
DocScrutinizerand I suggest you don't try to implement stuff that 5 other devels discussed for months and concluded it's not feasible17:03
DocScrutinizeror, even when feasible, is absolutely not what we'd like to see n kernel domain17:05
DocScrutinizerPali: honestly, if you want to improve hostmode kernel, you have no option but read and understand *all* that's been reported and discussed in the 3 lengthy usb-hostmode threads17:08
DocScrutinizerthen if you feel you're smarter, go ahead and discuss with the original devels - or just dive in head first. But then don't come and complain17:08
DocScrutinizerI'm happy to share some of my links: ff, and of course the mailing list stuff on h-e-n garage (not much)17:13
*** int_ua has joined #maemo-ssu17:22
PaliDocScrutinizer, I have other problem with musb/?? with kernel-power: When I connect my n900 to notebook usb charging port n900 detect that charger is connected instead usb data17:30
Paliso I cannot use that usb port on notebook for n900 Pc-suite/mass storage17:30
Palibut with maemo pr1.3 kernel all is ok17:30
Paliwhere I can start to debug this problem?17:30
Paliusb charging port is classic port on my notebook, but it can charge phone if notebook is turned off too17:31
DocScrutinizercharger detect means 1707 PHY has detected a D+/- short17:31
DocScrutinizerso your PC port has some means or bug that triggers this detection in 1707PHY17:32
gulydsmetool -k /usr/bin/modest17:33
gulyProcess not killed: not found, not root or kill failed17:33
Palibut why with stock nokia pr1.3 kernel there is no problem?17:33
gulyi'm root17:33
gulyany hint ?17:33
Paliguly: killall modest17:34
gulydoesn't work, spawn again17:34
DocScrutinizerto detect D+/- short, the 1707 connects one D-line to VBUS via a "pullup" resistor, and enables pulldown resistor on the other D-line. Then it checks if the line with pulldown resistor is pulled up to VBUS level17:34
DocScrutinizerPali: a good question why it works on stock kernel17:35
int_uaguly: I've just renamed modest executable. You are trying to get rid of it, right?17:35
gulyi'm trying to use it just when i need it, then actually kill it17:35
DocScrutinizerseems stock kernel either handles config of 1707 differently, or it handles OTG (aka charger) IRQ line from 1707 differnetly17:35
merlin1991guly: might be17:35
merlin1991dsmetool -k /usr/bin/modest.launch17:36
merlin1991or however that launch binary was called17:36
gulytried, doesn't work too17:36
gulyusr/bin/modest -> /usr/bin/maemo-invoker17:36
gulykilling maemo-invoker scares me :)17:36
merlin1991dsmetool ought to have some way to dump it's current watchhandlers17:36
merlin1991yeah that would kill pretty much everything17:37
merlin1991phone-ui, browser, everything that has a booster process17:37
DocScrutinizerPali: I think freemangordon had a look into some similar issue around fastcharger detect, that caused problems with shutdown or startup of kernel when charger attached, or sth like that17:38
DocScrutinizerI'd guess it's same root cause17:38
NIN101guly: to get rid of modest, find modest.desktop and nuke the preload stuff...17:40
NIN101I mean the X-preload or something17:40
gulyNIN101: yes, commented and removed those two lines17:40
gulydoesn't work17:40
NIN101you need to reboot17:40
gulyyes i know :)17:41
DocScrutinizershould work17:41
NIN101well worked for me :-).17:41
gulyso must be something strange with my configuration..17:41
DocScrutinizeractually I had to do it (for xchat that I eventually had on "autostart" by that feature) in several places17:42
DocScrutinizerdrove me nuts until I removed it again17:42
*** DocScrutinizer51 has joined #maemo-ssu17:43
*** Jaffa has joined #maemo-ssu17:43
*** changes topic to "Maemo Community Seamless Software Update "CSSU" channel, | Known bugs: | Channel logs: | Sources: | Latest version (testing): 21.2011.38-1Tmaemo2.1; (stable): 21.2011.38-1Smaemo2.1"17:43
DocScrutinizeriirc there are several places where this stuff gets cached or might live - can't remember the details, but I clearly recall I thought "duh, it "doesn't work")17:43
merlin1991why did the irc server adjust the topic?17:43
DocScrutinizerno adjustment17:43
DocScrutinizerjust server restart17:43
merlin1991I've set it to include a metting next week at the end tonight17:44
DocScrutinizernetsplit or sth17:44
*** arcean_ has quit IRC17:44
merlin1991now it got set back17:44
*** arcean_ has joined #maemo-ssu17:44
*** ChanServ sets mode: +o merlin199117:44
DocScrutinizerthey had to restore a backup obviously ;-P17:44
DocScrutinizerbarjavel freaked out last night17:45
*** merlin1991 changes topic to "Maemo Community Seamless Software Update "CSSU" channel, | Known bugs: | Channel logs: | Sources: | Latest version (testing): 21.2011.38-1Tmaemo2.1; (stable): 21.2011.38-1Smaemo2.1 | meeting: 28.03 20:00 UTC"17:45
*** merlin1991 sets mode: -o merlin199117:45
DocScrutinizer[2012-03-24 16:43:03] [Notify] Jaffa ist online (
DocScrutinizer[2012-03-24 16:43:03] [Notify] DocScrutinizer51 ist online (
DocScrutinizer[2012-03-24 16:43:04] [topic] #maemo-ssu Maemo Community Seamless Software Update "CSSU" channel, h17:46
DocScrutinizernfc if there's sth like a home irc server for a channel17:47
gulyactually it works, i must stop drinking aloe-based drinks17:48
DocScrutinizeranyway DocScrutinizer51 been on barjavel last night, and acted up like mad17:48
DocScrutinizerguly: toldya :-D17:48
DocScrutinizer[2012-03-24 16:48:43] [Whois] DocScrutinizer51 ist auf dem Server (Paris, FR).17:48
gulymodest has some issue with true-crypted gconf dir but this is another story :)17:48
DocScrutinizermerlin1991: maybe the "master record" for this chan's /topic lives on barjavel17:49
NIN101hmm truecrypted gconf dir17:49
NIN101good idea actually.17:50
*** trumee is now known as trumee_afk17:50
merlin1991I don't trust truecrypt :D17:50
NIN101yeah I am moving to cryptsetup (LUKS)...17:51
DocScrutinizeronly if truecrypt passphrase requester doesn'T need *any* service that's using gconf keys17:51
*** trumee_afk is now known as trumee17:51
gulywell i encrypted just apps/modest dir17:52
NIN101what is the "apps" dir?17:52
guly /var/lib/gconf/apps/modest17:52
gulyhave to go, bye17:53
DocScrutinizerI doubt you can simply add gconf dirs later on - I'd expect gconfd will try to access all of them on startup17:53
DocScrutinizeranyway, the day fades away17:54
DocScrutinizer(wow, pure lyrics)17:54
DocScrutinizergotta enjoy the big bluebox17:55
*** BCMM has joined #maemo-ssu17:59
*** ruskie has quit IRC18:00
*** int_ua has quit IRC18:01
*** ekze has quit IRC18:11
*** ekze has joined #maemo-ssu18:12
*** trumee is now known as trumee_afk18:18
*** BCMM has quit IRC18:19
*** ruskie has joined #maemo-ssu18:22
JaffaDocScrutinizer: ?18:36
DocScrutinizersorry for highlighting you18:37
JaffaAh, wasn't specific?18:37
DocScrutinizernope, just to document the returning barjavel irc server18:37
DocScrutinizeranyway, hi! :-D18:38
DocScrutinizerJaffa: barjavel reset the chan /topic to an obsolete version18:40
DocScrutinizerstrange thing18:40
JaffaDocScrutinizer: Heh18:41
JaffaDocScrutinizer: hello o/18:41
*** javispedro has joined #maemo-ssu18:45
*** andre__ has quit IRC18:52
amiconnDocScrutinizer: Two packages which got lost with earlier cssu versions are fixed now (rtcom-accounts-voip-support and mdest-home-applet). The one which still gets lost is microb-geolocation.19:32
amiconnmerlin1991 too ^19:34
*** sam23m has joined #maemo-ssu19:46
sam23many chance bug 11976 will be included in cssu?19:47
povbotBug viewing sms conversations in portrait mode without horizontal scrolling (patch)19:47
*** Pali has quit IRC19:48
merlin1991sam23m: this one is quite controversal19:48
merlin1991we can't reship the package due to the license19:49
merlin1991so we have to build a proper patch pacakge19:49
merlin1991like it is proposed in the last entry by pali19:51
merlin1991though nobody ever invested the time to actually build such a thing19:51
*** trbs has joined #maemo-ssu19:53
DocScrutinizermerlin1991: eh? the src is open, but licence forbids reshipping?19:55
merlin1991something like that if I remember correctly19:56
merlin1991also src is open because the part to fix is a css file19:56
merlin1991you can't close source that stuff ;)19:56
DocScrutinizerI'd not give a flying F on (C) of a CSS19:57
DocScrutinizerhonestly if you got any problem with that, ship a patch instead ;-D19:57
merlin1991that is a problem by itself19:57
merlin1991there is no "patch" utility in fremantle19:57
merlin1991only in the sdk repo19:57
merlin1991though it should be possible todo it19:58
DocScrutinizeranyway, patch is GPL and a simple cmdline tool19:58
merlin1991first of that19:58
DocScrutinizerno problem to ship it with CSSU19:58
merlin1991and secondly since we can onsider rtcom fixed in version a simple echo >> cssfile should work aswell :P19:58
DocScrutinizerumm, you only need to append lines to "fix" the issue? great!19:59
merlin1991yeah the only thing to consider is proper dpkg-divert or something to make it sane for a possible update20:00
DocScrutinizeranyway iirc diff can create several different output formats. one of them might fit for messybox sed or sth20:00
merlin1991in other words someone has to spend about 2 hours medling with packaging untill it is secure and works20:01
DocScrutinizernot me, I hate friggin portrait mode20:01
merlin1991I think the best approach would be to have a package that depends on the rtcom package, has a md5sum compare check for the version20:02
merlin1991and in case of "yes we can fix this" applies the echo >>20:02
merlin1991everything else should be taken care of by dpkg then20:02
merlin1991I just need to figure out how dpkg-divert works for real20:03
merlin1991it's a bit of black magic to me since I never looked into it20:03
DocScrutinizerdpkg is all black magic to me20:03
DocScrutinizeras I never really looked into *any* packaging20:04
DocScrutinizerexcept installshield ;-P20:04
DocScrutinizerdang, that been a P20:05
merlin1991hehe installshield, kinda as bad as the native ms installer stuff20:05
* DocScrutinizer still throws up, after 12 years20:05
sam23mhow about this one
DocScrutinizermake sure it works on win98, 98se, ME, NT, 200020:07
sam23mnot sure if there is a bug report somewhere20:07
DocScrutinizerreally have to run now20:08
merlin1991sam23m: afaik, it never was found out *why* the database goes corrupt, and afaik the involved components are closed source20:08
sam23moh thanks!20:08
merlin1991so far I've only seen a "fix" to recover a corrupted databas20:09
DocScrutinizernota bene structure of eventsdb changed IIRC with PR1.220:27
DocScrutinizerso it might already have been fixed20:27
sam23mi still experience it20:28
DocScrutinizeranyway, like for all messaging apps (mail, sms, whatnot else) it might be reasonable to move really old stuff out of the online database and keep them in some archive20:29
DocScrutinizerperformance of any database doesn't improve with a large number of records20:30
DocScrutinizerbtw!!!!! this seems to me like a plausible root cause for performance issues due to excessive ram usage and subsequent swapping: sql database size too large20:31
DocScrutinizerthis might explain why some users seem to suffer frequently and terribly from such sluggishness of system, while others like me (with really small sql database) never really encounter it20:32
DocScrutinizerDIE TRACKER DIE!!!!20:32
merlin1991yeah I only have problems with slow device when tracker goes mad20:32
DocScrutinizerit'S been the absolutely worst design decision ever to user trackerdb a lot for core things20:33
merlin1991or if I start openoffice in easy-debian, but that's bound to happen :D20:33
merlin1991tracker isn't that bad, it's bad how it's used in harm20:33
merlin1991on maemo it does what it should20:33
merlin1991index the device media for the media framework20:33
DocScrutinizerit also is used to store sms etc, no?20:33
merlin1991sms are in a seperate sqlite db20:34
merlin1991it stores sms and email on harm20:34
DocScrutinizerok, then that'S applicable only for HARM20:34
merlin1991yeah tracker on maemo just needs a lil saner filter options (see album art in music files) + less I need to reindex everything now because I feel like then it would perfectly do its job20:35
DocScrutinizeranyway sql engine tends to allocate SHITLOADS of RAM, and that happens for eventsdb on any inbound our outbound call or SMS, for tracker with any change on the monitored dirs, etc20:35
DocScrutinizerwhere $SHITLOAD_of _RAM is proportional to size * number of records in db20:36
DocScrutinizerso no suprise it eventually brings system to a grinding halt, no?20:37
DocScrutinizersame for mails I guess20:37
DocScrutinizerI feel like finally spotting the king of bugs after 2 years of hunting ;-D20:39
DocScrutinizerquite obviously Nokia never tested their system performance with shitloads of records in those databases. Maybe for tracker with music, but even with gallery they for sure didn't20:41
DocScrutinizerand you bet they haven't ever even thought about an eventsdb with 15000 SMS in it20:41
DocScrutinizerand 20000 phoncall events20:42
DocScrutinizerwe should start a new investigation project for this20:42
DocScrutinizer"the mystery of sluggishness exposed"20:43
*** sam23m has quit IRC20:46
*** Pali has joined #maemo-ssu20:54
*** nox- has joined #maemo-ssu21:09
amiconn15000 SMS? o.O22:04
*** javispedro has quit IRC22:16
*** trumee_afk is now known as trumee22:16
*** bindi has quit IRC23:15

Generated by 2.13.0 by Marius Gedminas - find it at!