/srv/irclogs.ubuntu.com/2008/06/03/#ubuntu-mobile.txt

asaclool: how can we roll -updates for diverged packages to hardy PPA?00:03
asacdo we need some official staging area to get some QA first?00:03
=== Moot2 is now known as MootBot
persiagood morning01:00
=== asac_ is now known as asac
dholbachgood morning07:58
=== emgent_ is now known as emgent
loolasac: The staging area is going to be the ppa10:33
loolasac: And we will then promote updates to the mobile archive10:34
persialool: I'm looking at bugs 215842 and 218251.  In the latter, you recommend against my solution for the former.10:36
persiaAre you sure we don't want a conffile for that?  I prefer conffiles as they can be purged, whereas maintainer-script generated files tend to stick around...10:36
loolpersia: hahaha, I was a good guy one month ago10:37
loolpersia: That comment was just to make sure that /etc/default/locale shouldn't be in a package10:37
loolAnd I was bitten by this no longer than two days ago, because it's shipped in ume-config-common10:38
persiaIs it now?  Then why is the bug with my patch still open?  If that patch was applied, it ought be closed.10:38
loolpersia: if you run simple-vm-builder, it will hand waiting for a merge during the vm creation10:38
looldpkg -L ume-config-common| grep locale10:39
lool/etc/default/locale10:39
lool  * Add a default locale for builds to support UTF8, etc. Patch by Emmet10:39
lool    Hikory.10:39
lool -- Steve Kowalik <stevenk@ubuntu.com>  Thu, 29 May 2008 18:20:32 +100010:39
persiaThat's why I don't understand why 215842 is still open.  That's from where the patch comes.10:39
loolProbably an omission10:40
loolpersia: Would you like to fix shipping of /etc/default/locale?10:40
loolWhat should happen is more something in the postinst like: if /etc/default/locale exists and sets a locale, leave it alone, otherwise create or set the locale10:41
persialool: Actually, I prefer it as a conffile: I just encountered your recommendation against it being a conffile, and thought it might be worth discussing in case we want to change it before we ship.10:42
persiaOK.  How do we unset the locale later then?10:42
persia(on package removal)?10:42
loolpersia: It can't be handled like a conffile in UME and not anywhere else10:42
loolpersia: Not sure whether we want to unset the locale10:42
persiaOK.  I can see the case for not unsetting it.  I'll push a fresh config with maintainer-script handling then.10:43
loolthanks10:43
loolpersia: And if you like, you can verify that it breaks e.g. simple-vm-builder if you're not convinced about not shipping it as a conffile?10:44
loolIt would also break if you'd install ume-config-common on e.g. a regular system10:44
persiaIt oughtn't break, just cause conffile-collision handling.10:45
loolYes; but it's supposed to provide either defaults or overrides, not change the way a file is handled  :-/10:45
persialool: Right.  I'll fix it in the next hours :)10:46
loolUltimately, I suppose we're missing an "install" step where such a file would be created, just like /etc/hosts; this could be done with the current install.sh hacks10:49
persiaDo you think that is better than in ume-config-common?10:50
StevenKPersonally, I think ume-config-common is a bad idea, and stuff it does should be handled on install10:51
persia(You say "ultimately", but I read it as "before this upcoming release")10:51
persiaLet's do it on install then.  Where do we need to stuff the hook?10:51
loolpersia: I meant in a future cycle / world10:52
loolpersia: Doing it on install would mean rerolling a MIC; we could do it as it's still planned to move to the new sources.list10:53
persialool: Right.  My worry is that if we do it wrong now, we're somewhat stuck.  There's no purge part during the upgrade, and unmarking conffiles is an ugly hack.10:53
persiaStevenK: How are we on rerolling a MIC?10:53
loolYes, I agree it's ugly to fix this later on10:53
loolpersia: If you do it in install.sh, be careful that it might busybox sh10:54
loolI think u-v-b has a good shell snippet of what's needed, but it might need to happen at the end of install10:55
* persia needs guidance on where to put stuff for install.sh, but isn't worried about busybox vs. bash10:55
loolpersia: Grab MIC10:56
loolplatforms/common-apt has the initramfs stuff10:56
persiaRight.  MIC.  Thanks.10:57
loolinstall.sh is called from platforms/common-apt/initramfs/usb and /cd if present in the initramfs10:57
=== ogra_ is now known as ogra
ograpersia, /etc/default/locale gets rewritten by gui tools on locale changes11:03
ograa conffile wouldnt be clever here11:03
persiaogra: Right, which is why it is being removed :)  On the other hand, those GUI tools aren't in UME right now (see bug 218251)11:04
ograbut they might be in the future 11:04
persia-ENOBUGBOT https://bugs.launchpad.net/ubuntu-mobile/+bug/21825111:04
persiaogra: Right.11:05
ograi.e. in a netbook image there should be as many desktop tolls from the defaut install as possible imho ... else we have to maintain everything twice11:05
ogra*tools11:05
persiaogra: Which reminds me, have you put together a meta for your netbook interface idea?11:06
ograi.e. we should at least use the same backends ....11:06
loolpersia: The GUI tool is in customer builds though11:06
* persia wants to test on the Kohjinsha, as neither Mobile nor Ubuntu is a good fit11:06
ograwell, the meta is ubuntu-desktop ... i need to put together a theme and gconf settings 11:06
ograthe app selection is 100% ubuntu-desktop yet11:06
persiaogra: Just gconf settings?  That's enough to change the interface?  Nothing new is required?11:07
ograah, well, devilspie to make metacity behave like matchbox ... and a two line devilspie config 11:07
ograand one little script hack to get the close button in the panel 11:08
ograut the rest is gnome ui changes11:08
ogra*but11:08
* persia imagines a seed depending on Ubuntu-Desktop and containing a single -settings package11:08
ograwell, -sessings and -theme 11:08
ogrand yes, that was exctly the idea :)11:08
ografind the apps in ubuntu that misbehave and add proper gconf keys .... and have the world fixed in one go :)11:09
davmor2Guys just to let you know the KVM version of RC1 is more reliable than the mic version11:09
persiaDoes anyone have a ume-config-common-0.12-1-0ubuntu0 source package?11:09
ograno desktop app maintenance apart from the fixes and patches which might even be able to go upstream11:09
persiaUpstream is good :)11:10
ogra:)11:10
StevenKpersia: I have it locally, I think11:13
persiaStevenK: Did it exist anywhere else?  Can we get it back, or do we need to push -2?  (Assuming that few people are expecting continuous upgrade to work at this point)11:15
loolpersia: -0ubuntu0?11:17
persialool: The PPA gave me a diff file exactly matching that which I want to revert against -0ubuntu011:18
loolpersia: The PPA has indeed11:18
loolhttps://launchpad.net/~ubuntu-mobile/+archive?field.name_filter=ume-config-common&field.status_filter=any11:18
lool(0.12-1-0ubuntu0)11:18
persiaAha!  Now, to delete the new one (and save uploading a paper bag)...11:19
StevenKHah11:20
StevenKpersia: So you don't need me to find it?11:20
persiaStevenK: Nope.  lool knows LP well enough to save us both the effort :)11:20
StevenKpersia: Heh, fine11:21
=== cprov is now known as cprov-lunch
persiaThe PPA archive-admin stuff isn't so bad, really.11:34
=== davmor2 is now known as davmor2_away
cprov-lunchpersia: ehe, I will take it as a compliment ;)11:35
* dholbach hugs cprov-lunch11:36
persiacprov-lunch: If you're responsible, you ought :)11:38
loolpersia: cprov-lunch beeps on "ppa", guess how responsible he is :)11:43
* persia ponders the difference for being responsible for a thing and being responsible for creating a thing11:44
* persia seeks peer review of http://paste.ubuntu.com/16558/ as a solution for /etc/default/locale before committing11:58
ograpersia, does that check if the file exists and is pre-set somewhere already ? 12:14
ogra(your patch doesnt seem to)12:15
asaclool: are you on amd64?12:16
persiaogra: Nope.  This is run on the initialisation of the filesystem, prior to any packages.12:16
ograah12:16
persiaIs that safe?12:16
ogralooks fine imho 12:16
ograwell, thats what i do in all my chroots as well ... does it run locale-gen soewhere with the matchng value ? 12:17
ogra(it wont be pregenerated nowadays)12:17
persiaHrm.  I'll have to look for that.  In previous testing, the mere existence of /etc/default/locale was enough to make most applications behave (as in use UTF8 rather than POSIX)12:24
ograi'm not sure which package does the locale-gen today ... it used t be done in the locale package, then moved to language packs but might be in language-support now which you likely dont want in default mobile as it pulls in tons of extra stuff you dont need12:25
persiaogra: Yeah.  I think I'll just stuff in this file for now, as I believe pushing it at filesystem-creation time is better than pushing it at ume-common-config time, and has fewer negative implications.12:28
persiaTracking down locale-gen can be done another day.12:29
ograyeah12:29
cprov-lunchpersia: thanks (and yes, I try to be responsible in both sense)12:40
=== cprov-lunch is now known as cprov
=== persia_ is now known as persia
loolasac: I have an amd64 box13:32
asaclool: ok, then the libflashsupport problem is not that bad (referring to mail)13:34
loolasac: ogra questionned by 64-bits use as well13:34
ograasac, what makes me curious is that lool says it works fine on debian with the debian version13:35
ogra(32 bit here)13:35
asacno pulse?13:36
asacdebian version of what?13:36
ogralibflashsupport withour pulse ? would that make sense ? 13:36
ogralibflashsupport13:36
ogra<lool> ogra: I'm using a custom install of libflashsupport on Debian/i386 (1.0~2219-1) and 1.9-0ubuntu1 on Ubuntu/amd6413:36
ografrom -devel ^^^13:36
asaclool: a custom install?13:38
asaccan you post the build config for that please?13:38
asaclool: what kind of system spec does the debian system you test on have?13:39
lool(hold on, phone)13:44
=== davmor2_away is now known as davmor2
ian_brasili am trying to add an icon to the grid desktop and the info on https://help.ubuntu.com/community/UMEGuide/ApplicationDevelopment/AddingIconToDesktop does not seem to work13:58
ian_brasilany ideas?13:58
persiaian_brasil: Did you add OnlyShowIn=GNOME;Mobile; ?13:59
ian_brasilno...where do i add this?13:59
StevenKIn the .desktop file13:59
persiaIn the ,desktop file.13:59
ian_brasilah ok13:59
persiaAlso, I'd recommend just putting it in /usr/share/applications, unless you have a good reason to do otherwise.14:00
persiaUbuntu Mobile no longer only looks in /usr/share/mobile-basic-flash/applications/14:00
ian_brasilok..i will update the wiki 14:01
persiaian_brasil: Thanks.14:01
ian_brasilso it looks in /usr/share/application first ..then /usr/share/mobile-basic-flash/applications ...then anywaher else?14:06
StevenKIt looks in /usr/share/applications only14:06
persia(and subdirectories thereof)14:06
loolasac: I'm using the universe flashsupport on Ubuntu/amd6414:07
loolThe custom install probably comes from the website of flashsupport for the Debian install14:08
asaclool: ok. we should figure their build config then. what version?14:08
lool1.0~2219-114:09
ograis that packaged already ? 14:09
ogra(in the archive i mean)14:10
loolYes, not the smae version14:10
ograit wasnt even in experimental and not clear if it would go in when i looked last time14:10
loolOn Ubuntu, I have 1.9-0ubuntu114:10
ograand it ships a hard openssl dep iirc14:10
loollibflashsupport | 1.9-0ubuntu2 | intrepid/universe | source, amd64, i38614:10
ograright14:11
ograthe code should be identical though ... both pulled from pulse git which didnt change since a year or so14:11
asaclool: where did you get that custom package from?14:11
ograthere is an alioth repo14:12
loolHowever there's also a new flashplugin-nonfree-extrasound14:12
loolSorry, I mean flashplugin-nonfree-pulse for pulse14:12
ograyeah, seems they renamed it14:12
loolIt's split actually14:13
loolI can't find flashplugin-nonfree-pulse though; only flashplugin-nonfree-extrasound which refers to it14:13
asaclool: how do you test?14:13
ograhttp://git.debian.org/?p=pkg-pulseaudio/flashplugin-nonfree-pulse.git14:13
asaclool: the test case is going back and forward on youtube until flash/firefox crashes (if you dont have nspluginwrapper)14:13
loolasac: Nothing in particular, I open flash videos14:13
asaclool: well. try the case above14:13
asacgo to youtube video, wait for movie to start, go to new youtube video, wait for start, go back + wait, go forward + wait14:14
asacif you dont have nspluginwrapper in between + you are sure libflashsupport is loaded (e.g. strace) and you are sure that pulse is actually used by flash, then we might have a fix :)14:14
ograi wonder how that should have magically appeared though14:15
loolasac: how many times?14:15
asacremember that crimsum uploaded some workaround/fixes to intrepid. maybe those are already in debian now.14:15
ogralookig at that git repo there are no changes to lennarts code at all14:15
ograasac, that was to get rid of the fixed openssl dep ....14:16
asaclool: no idea. i never made it to 10 retries14:16
ogranothing that would affect non encrypted playback14:16
loolI see flashplayer in pavucontrol14:16
asaclool: but the slower your sytem is the less likely you will hit the deadlock14:16
asacogra: well ... crimsun also claims that he fixed the crashes in his upload14:16
loolI certainly crash my system from time to time on flash videos14:16
ograasac, oh14:16
loolmy browser I mean14:16
asacogra: i pinged him multiple times for the last two week to get input on what he did, but he closed all bugs14:17
asacnever got a reply though14:17
ograhmm14:17
asaclool: try the testcase above ... if that doesnt crash your browser then its a good sign14:17
asacogra: remember that crimsun had a fix for it in hardy on also side, however that wasn't regression free.14:18
asacmaybe he uploaded to intrepid in the hope that those can be fixed during cycle14:18
ograthe upload i see in intrepid only changes one dep14:18
asacs/also side/alsa side/14:18
ogra* debian/control:  Conflict with flashplugin-nonfree versions older14:18
ogra     than intrepid's while allowing for backports (LP: #192888).14:18
asacogra: yes, but i think that he uploaded other packages in that batch14:18
ogra   * New upstream beta fixes many crashers (most significantly14:19
ogra     LP: #192888).14:19
ograflash itself14:19
asacyeah ... i could reproduce the crashes with flash 10 ... the version i tried at UDS14:19
asacso that alone cant be it ... unless the current upload is more recent and indeed fixes things14:19
asaclool: using flash 10 from intrepid?14:19
loolNo14:20
asacand zero crashes?14:20
lool9.0.124.0ubuntu214:20
asac(when doing the test?)14:20
loolCertainly not zero crashes in general, but I've hit a least 8 times back/forward in youtube now14:21
loolOn the same video14:21
asacnot that bad14:21
loolNote: one of my CPU is at 100%; it's a fast system, but could behave like a slow system in this case14:21
asaclool: can you reproduce the crashes on ubuntu? maybe respin the debian libflashsupport on ubuntu?14:22
asaclool: building?14:22
loolNot building, just playing the video14:22
loolIt's the same on both computers since forever with flash14:22
asacsure you are using flashplugin nonfree?14:22
asacstrnage14:22
loolI got a crash the very first time I connected to youtube today, but since that...14:23
ian_brasilis there some way to turn icons on or off?...like i want cheese icon to appear in gnome but not in hildon14:23
persiaian_brasil: The OnlyShowIn key is the best mechanism to control that.14:24
ograbut will switch the item on/off completely14:24
persiaogra: Selectively, so you can show or not show in e.g. GNOME or Hildon14:24
ograthe xdg spec could need an addition14:25
loolasac: [pid 21542] open("/usr/lib/flashplugin-nonfree/libflashplayer.so", O_RDONLY) = 314:25
lool[pid 21542] open("/usr/lib32/libflashsupport.so", O_RDONLY) = 3514:25
ograpersia, yes, but ian_brasil wants to switch visibility of the icon without losing the entry i guess14:25
ian_brasilright14:25
asaclool: looks like nspluginwrapper?14:25
ograthe xdg spec doesnt provde such a feature14:25
persiaHrm?  How is a menu supposed to show an invisible icon?14:25
asacthat will guard you from crashes14:25
asaclool: check whether npviewer is running while playing flash14:26
ograit will falll back to the default icon in gnome 14:26
ogranot sure hildon has such stuff14:26
persiaian_brasil: Could you summarise this use case?  I'm fairly familiar with the specs, but a little tired now...14:26
ian_brasilok...i will mail the list14:27
loolasac: Hmm pavucontrol is failing on the amd64 box; probably due to old pa env vars, but new pa running14:28
loolasac: Oh you mean I should remove nspluginwrapper?14:28
asaclool: well ... if nspluginwrapper is used you wont see the crashes ;)14:29
asacso on 64-bit its not an issue for us14:29
asacnot sure about debian, but if you ar e on 64-bit there too then you wont see the bug14:29
ograand on i386 we'll catch it the same way in intrepid :)14:29
loolI'm on i386 on Debian14:29
asachopefully14:29
asaclool: maybe debian has nspluginwrapper on i386?14:30
loolSo I'm sorry guys, I'm a bit lost in what you wanted me to test14:30
ograflash without nspluginwrapper on a ubuntu system using pulse :)14:30
loolI just know I have been using libflashsupport (heck, I even hacked it for my previous job to make flash decode audio for us :-) but what crashes other people I don't know14:30
ograpreferably i38614:30
loolOk; I tested that on Debian only14:30
ograright, if debian doesnt use nspluginwrapper that sounds  like a fix for us ...14:31
asaclool: i just wanted to know whether you have nspluginwrapper on your debian install14:31
ograbut we dont seem to differ in code at all14:31
ograwhich is weird14:31
loolI don't have nspluginwrapper on my Debian install14:32
asacbecause its strange that you dont see the crash14:32
asaclool: but you installed flashplugin-nonfree from somewhere ... maybe that includes nspluginwrapper?14:32
asacif no npviewer is running then you probably dont use it. and we should see why you dont get issues14:32
loolOn Debian [pid 27823] open("/usr/lib/flashplugin-nonfree/libflashplayer.so", O_RDONLY) = 6014:33
lool[pid 27823] open("/usr/lib/libflashsupport.so", O_RDONLY) = 5014:33
asaclool: you have a link to the debian source package you use?14:33
loolI don't see any process with np14:34
asack14:34
loolI use libflashsupport (1.0~2219-1) unstable; urgency=low14:34
lool -- CJ van den Berg <cj@vdbonline.com>  Tue, 28 Nov 2006 12:29:54 +010014:34
loolI think I got it there http://pulseaudio.revolutionlinux.com/PulseAudio14:35
loolAh http://pulseaudio.vdbonline.net/libflashsupport/ dropped14:35
loolhttp://pulseaudio.vdbonline.net/ has flashplugin-nonfree-pulse_0.1~000.dsc14:35
asachmm14:37
asaci think i am too tired ;)14:38
loolI think the pulse impl was rewritten at some point, but IMBW14:38
ograit was14:40
ograby lennart -... back when he still was DD 14:40
ograbut he didnt touch it since14:40
loolHe isn't DD anymore?14:41
ograand now he works for RH keeping him busy with other stuff 14:41
ograhe still is, but makes no use of it14:41
davmor2pulse and avahi if I remember correctly isn't it ogra14:58
ograthats what he did in debian, yes14:58
ogranow he works for RH, does rpm's and doesnt care much about debian packaging anymore14:58
ograwe talked a bunch in prague 14:59
davmor2:(14:59
ian_brasilthis is intersting -> cheese does not have OnlyShowIn but it appears in hildon ...others also do not have OnlyShowIn and do not appear..seems like some default list14:59
ograian_brasil, its always more than just the .desktop files, the .menu files have impact as well on what is shown and what not15:00
persiaogra: There's a bit of a specialty implementation for Ubuntu Mobile...15:00
ograi.e. if you disable an item in alacrate in ubuntus normal desktop it wont change the NoDisplay value of the .,desktop file but add a ~/.local/menus/*.menu file ecluding it15:01
ograwell, and then there are mobile hacks as well apparently :)15:01
emgentheya15:04
ograerr, sorry, its ~.config/menus/15:04
ian_brasilanother queston..i want to write a status panel plugin in python to shutdown the system..this is possible right15:36
ogratricky if you dont use the default desktop ... i.e. dont have a gdm socket to send the signals to15:38
persiaian_brasil: For what conceivable reason would you ever want to shut down a MID?  Aside from system crashes, I've never turned off any of mine for the past several years.15:38
persiaHibernate is the key, and setting a short sleep timer.15:38
ograor suspend :)15:38
ograhave a look at pm-utils for that 15:39
ograpm-hibernate and pm-suspend specifically15:39
ian_brasilit is not a MID i am developing for!15:40
persiaAha!  In that case, yes, just hook into pm-utils :)15:41
ograwell, for shutdown/reboot you either need a DM to do it proper or add evil hacks to sudores or so15:41
ograpm-utils doesnt provide hooks for shutdown/reboot iirc15:42
ian_brasilok thx15:43
ografor gdm you would want: gdm-signal -h or -r15:44
ograassuming you have gdm running with autologin15:44
ian_brasili want to do something like gnome with turn off reboot, suspend, hibernate15:44
ograwhy not use the gnome one then ? ... 15:45
ograoh, wait you dont use gnome-panel/-session i guess15:46
ograwhere the dislog is implemented15:46
ogra*dialog15:46
ian_brasili am using gdm :)15:46
ograits easy to en/disable features in that dialog through gconf keys andchanges to gdm.conf15:46
ograwell, thats a start, try with the gdm-signal command15:47
ian_brasilok15:47
ograit also offers interfaces to suspend/resume i think (calling pm-utils in the backend)15:47
ian_brasilgdm-signal does not work..not even with sudo15:48
ian_brasili will reboot and try again15:49
ian_brasilgdm-signal does not seem to work...does not return anything15:54
ograhmm15:56
ograi know thats what gnome-session uses internally15:57
ogragnome-session/gdm-logout-action.c and gnome-session/logout.c use it extensively15:58
ogra(directly though, not through the gdm-signal binary)15:58
ian_brasili tried with xfce too and that does not work either15:59
ograhmm16:03
ogralooks like it doesnt work here either 16:03
ograit comes from powermanagement-interface which isnt really maintained anymore (just as compatibility layer for pm-utils)16:04
ograhmm, strace shows its using wrong strings to send to gdm 16:26
ograhttp://paste.ubuntu.com/16599/16:26
=== cprov is now known as cprov-out
ograian_brasil, hey, the trick is to end your session, gdm-signal only queues your comand until the session has ended, then gdm executes it17:21
ian_brasilogra: cool..i'll try that17:29
ograi just found that out by accident when my system suspended while i had only logged out :)17:30
aileanguys, can anyone help me choose an ubuntu-friendly smartphone/pda? I'm not overly concerned about running ubuntu/linux ON it (although that would be nice), just to be able to use it with my ubuntu laptop well.18:16
=== Sciri is now known as Guest30607
=== Sciri_ is now known as Sciri
bspenceranyone tried to use yelp in UME?23:12

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!