/srv/irclogs.ubuntu.com/2008/06/12/#ubuntu-mobile.txt

=== jldugger-tablet is now known as pwnguin
=== pwnguin is now known as jldugger-tablet
djp_hi ubuntu-mobile maniacs !16:18
djp_is there yet a channel for Remix ?16:18
=== djp_ is now known as djp
djpor is this the place ?16:21
looldjp: We don't have a channel for remix16:42
looldjp: I suspect remix people are around16:42
loolAt least njpatel is :)16:43
djpa couple of things I wodering;16:57
djpwill it be available for non-Atom CPUs ?16:58
djpwhen the anndouncement mentiones "optimized for retail" what do they mean ?16:58
djpto put on machines to sell ? or to put on machines for sales (I want the latter)16:59
loolMeeting in one minute16:59
djpwhen will it be ready ?16:59
djpok16:59
lool#startmeeting17:00
MootBotMeeting started at 11:02. The chair is lool.17:00
MootBotCommands Available: [TOPIC], [IDEA], [ACTION], [AGREED], [LINK], [VOTE]17:00
loolHi everybody; thanks for attending this week's meeting17:00
loolThe agenda is at https://wiki.ubuntu.com/MobileAndEmbedded/Meeting/2008061217:00
loolI love this week's agenda17:00
loolAs I read it, we have no items to revisit from last week17:00
loolNor for this week17:01
loolSo it's open agenda season!17:01
loolI'll hang around for 5 minutes or so in case anybody would like to add an agenda item17:01
cgreganshould we discuss RC3?17:01
lool[topic] RC3 discussion17:01
MootBotNew Topic:  RC3 discussion 17:01
loolcgregan: Sure, go ahead!17:01
* persia likes RC3: works better than any previous images17:02
cgreganI just wanted to say that is was built and is working great17:02
cgreganPassed all smoke tests17:02
GrueMasterwhere's rc2?17:02
loolCool; same here, it works great for me17:02
cgreganNot bug free....but I'd agree with lool and persia17:02
cgreganRC2 is still up in the release area on cdimage Gruemaster17:03
GrueMasterNo one notified us that it was ready.17:03
loolGrueMaster: I still see rc2 http://cdimage.ubuntu.com/moblin/releases/hardy/17:03
GrueMasterYea, I see it now too.17:03
loolHowever StevenK plans some cleanup of at least the beta images soonish17:03
GrueMasterno problem there.17:03
cgreganRC3 is much better GrueMaster17:03
cgreganWe believe this will become Final soon17:04
GrueMasterWill get and start testing it.17:04
cgreganThere are some great VM image building instructions over with the cases on the wiki17:04
cgreganhttps://wiki.ubuntu.com/Testing/Cases/UMEinstall17:05
cgreganThat's all I had on RC317:06
loolOk17:06
loolI'm currently spending my time on misc small bug fixes and on virtual-mobile-builder17:06
cgreganWe should develop some method of alerting the community to new builds17:06
cgreganCan we have an action item for me?17:07
loolcgregan: I think it will make more sense when we actually spend the time to build more generic images; e.g. targetted at eepc17:07
loolcgregan: Also, marketing announces releases here17:07
cgreganAhh17:07
loolcgregan: That said, I think you can happily take an action to talk to marketing to send announces about RC-N releases or betas17:07
loolIn fact, these should be announced along the regular Ubuntu releases in the next cycle IMO17:08
lool(alpha 1, 2, 3, etc.)17:08
cgreganWill do17:08
lool[action] cgregan to talk to marketing about announcing pre-releases of UME17:08
MootBotACTION received:  cgregan to talk to marketing about announcing pre-releases of UME 17:08
loolOk; if we're done with rc3, any other topic for today?17:08
persiacgregan: Might be worth also working with stgraber to get these as part of the ISO tracker (although these are not ISO images).17:09
cgreganYes...there is already discussion on moving testing to ISO tracker17:09
* persia likes alignment of effort17:09
loolOk; I see no new topic on the wiki page, so I'll just take this occasion to warn people that they should be very very careful about uploads to the ppa17:10
loolThe ppa is now our stable-updates testing area; once an update is tested, it will be promoted to the release archive17:10
persialool: That's now for SRU candidates only, right?  New stuff should go into intrepid?17:11
loolpersia: Exactly17:11
loolNo new development or releases should happen in the hardy ppa; unless needed to fix a grave stable bug, but that's unlikely17:11
* persia is willing to be a contact point for anyone having issues getting updates into intrepid (not always as uploader, but for process questions, etc.)17:11
loolGet in touch with me if you want to prepare such uploads and need guidance17:11
loolOr ask persia :)17:11
bspencerlool: there is also risk when image-creator changes.  I assume that can affect you too, tru?17:11
bspencere17:11
loolbspencer: Where would it change?17:12
loolbspencer: We currently use the packaged image-creator which is in our ppa17:12
bspencerfor example, it just changed to use autogen.sh 17:12
bspencerah, ok.17:12
bspencerno risk then17:12
loolWhich has the new platforms for the release archive17:12
loolThat new MIC should be targetted at intrepid now17:12
loolNot sure it was released though17:12
bspencerI haven'tused a packaged image-creator for over a yar17:12
bspenceryear17:12
loolbspencer: That explains :)17:13
bspencerall our docs say to pull from the source and build/install17:13
bspencerand that the packaged one is out of date and uncertain17:13
bspencer:-\17:13
loolbspencer: All our docs say to install the package ;)17:13
bspencercool.  glad it works actually17:13
bspencerI wasn't sure you were updating and testing with it.  Good news17:14
loolI can only use the packaged one17:14
loolIt has many fixes we rely on17:14
bspencerare the platforms in the packaged version highly customizes?17:14
bspencerd17:14
loolbspencer: "highly" not really17:14
loolWe add a couple of platforms17:14
persiabspencer: Sounds like a focus thing: it's likely best for upstream types to use VCS trunk directly, and for Ubuntu Mobile hackers to use the package, and push candidate changes in bug reports.17:14
bspencerwell, I mean have you eliminated unused ones.17:14
loolOne for mccaslin, one for menlow17:14
bspencerhardy+ppa17:14
bspencerand a new one for eeepc ?17:14
loolWe didn't eliminated uneeded ones17:14
loolThere's no platform for eeepc yet17:14
bspencerok.  thx.17:15
bspencerpersia: true, although we're all Ubuntu Mobile hackers (IMO)17:15
loolAnother thing we change in the platforms we build against is that we drop most packages in fsets and use the ubuntu-mobile metapackage to pull things instead17:15
bspencerlool: right17:15
persiabspencer: Excellent :), although I suspect we won't all also be upstream as we grow.17:16
loolbspencer: I'm happy to work on making the packages better17:16
loolbspencer: In fact I've prepared a non-negligible number of patches for MIC17:16
bspencerlool: great17:16
loolHalf of them aren't merged yet, so I wouldn't be able to move to a new MIC without rebasing them17:16
bspencerok.  We'll do that after your UME release17:16
bspencermeaning -- we can focus on getting those things in 17:17
loolThat said, I would recommend you to publish packages of any app you tell users to use; that is, if you're unhappy about the age of our packages, we could of course try to fix that, or you could publish newer MIC packages in other repos17:17
loolbecause manually installed apps tend to break subtly with old files not being removed and generally mess up one's system17:17
loolHow many times have I received bugs about crashes resulting from libs installed by the end user in /usr/local/lib  :-/17:18
loolAnyway, sorry, I'm getting offtopic :)17:18
bspencergot it17:18
loolI wish everybody a happy week then17:18
loolthanks for attending!17:18
lool#endmeeting17:18
MootBotMeeting finished at 11:21.17:18
looldjp: Back to you now17:18
looldjp: Concerning readyness, and availability; I would guess your questions are partly covered in the FAQ17:19
persiadjp: Regarding architecture: the majority of code isn't actually architecture specific, although you may have some testing to do in order to verify it works on an alternate architecture (not that there are that many SPARC or PPC mobiles about)17:19
looldjp: if not, I'd be happy to complete it17:19
loolpersia: technically, many bits are only enabled if arch == lpia though17:20
persialool: True, but some of those cases are bugs17:20
loolpersia: Yeah; it was a shortcut to do things like this17:21
loolAnd a nice shortcut, cause it is indeed much harder to do properly :)17:21
persiaVery much so.17:21
loolWhat do you people use to push .debs to your kvm?  you publish them over http and then wget them?  push to a ppa?17:22
persiascp?17:23
loolI like being able to simply scp them, but I'm not sure I can reach my kvm17:23
loolat least I can't ping its ipv4 from my desktop17:23
loolI guess a different network backend would help17:23
persiaHeh.  What are you using?  bridged?17:23
loolI think not; I'm using whatever is the default17:24
loolLet's try bridged17:24
looli think it didn't work on my laptop, but wifi is too special17:24
persiaAh.  usermode.  You want to add iface br0 to /etc/network/interfaces17:24
loolpersia: What do you pass kvm for bridged mode?17:25
persialool: https://help.ubuntu.com/community/KVM will probably answer your next couple questions as well.17:27
loolpersia: So I'm not too familiar with libvirt, but it seems either you create VMs without it and then configure each VM manually, or you create VMs with it, and then you need to use it17:31
loolDoes that mean we should be using libvirt in virtual-mobile-builder by default and generate a libvirt command-line?17:32
* lool needs to run out for a while17:34
djplool: thanks for the info and the support, I'm trying to get an exiting new point of sales system built for my caffe (all FOSS, of course) and Remix looks like it might be perfect for the 800*600 touchscreens17:34
persialool: I'm not sure.  I read that too, but am not awake enough to test much (having gotten up from a nap, and planning to sleep again in 85 minutes precludes coffee).17:34
persiadjp: If you're looking at it from a PoS perspective, you'll have a bit of customisation to do anyway.  I'd recommend starting with matchbox, and seeing how much you want from the remix, from Ubuntu Mobile, and from other sources.17:36
djplool: s/exiting/exciting17:36
djppersia: interesting.. matchbox, eh ?17:37
djpthe PoS app in question is a java/swing thing17:37
djpguess that wouldn't precluse matchbox17:37
djps/precluse/preclude17:37
djpbut I don't see it as a pure appliance, as there are several other apps usefull from the register17:38
persiadjp: Sure.  It's a matter of what you want to install.  Matchbox is just a window manager.17:41
persiaI've only limited experience working with PoS installations, but I suspect that except for the touchscreen interface, most of the use cases are quite different.17:41
persiaThere's no reason you can't have a panel of some sort, perhaps with launchers for your kiosk tasks, etc. (but this gets quite off topic)17:42
pmcgowanbspencer, hey, what is the issue with having shared use of xulrunner between browser and home plugin?18:00
loolpersia: I guess the page you sent me encourages moving to libvirt which will just do its things; I was happy to start vms with ./kvm-wrapper though; I also would like to keep network manager working on my ubuntu boxes, but I guess this conflicts with using bridges etc.18:05
loolvmwares manages to do it18:05
persialool: Yeah, well, it's a matter of software maturity.  You can still configure the bridged networking without libvert.  I'll dig you up another page.18:06
loolBasically no perfect solution on http://kvm.qumranet.com/kvmwiki/Networking18:10
loolThere's vde18:10
persialool: Are you on wireless?18:22
loolpersia: On the laptop, yes18:23
loolOn the desktop, no; would still prefer keeping NM though18:23
persialool: For the desktop, 3. public bridge from that page ought work, although it requires a bit of fiddling.18:23
loolThe slirpvde thing allows to do basically the same thing tahn kvm hmm18:23
loolpersia: Yeah, I'm used to such fiddling, but I know it's pretty much incompatible with NM then :-/18:24
persiaFor the laptop, you can create a fake eth0:1, and assign it some IP address, and use that (avahi-autoipd will do that for you: it's a bug for most people, but might be a feature for you)18:24
loolIt looks like we don't have a transparent solution like vmware's in opensource18:24
persiaWell, that's largely because we don't preload everything on the host.18:25
loolpersia: Hmm you can use an alias in place of a tap?!18:25
persiavmware installs a mess of special interfaces on the host18:25
loolThe kernel drivers behave differently18:25
persiaWell, yes, but there are ways around that.18:25
loolFor example the bridge will simply, well, be a bridge, but leave the IP address and all to your phys intf18:26
persiaIf we by default created 10 extra interfaces on the host for use by KVM, it would be easy.18:26
looland the nat doesn't rely on iptables AFAIK18:26
persiaYou don't use the alias in place of the tap, just in the brctl line.18:26
loolWell you still wouldn't get NM support if you had a bridge18:26
persiabrctl doesn't check link status, so as long as you can route, it works, and the kernel will shortcut rather than sending over the interface for localhost.18:27
loolUh you're really adding an alias interface to a bridge18:27
loolIt doesn't check it, it enforces it :)18:27
persiaWhat?  You don't get nm support on the bridged interface, but that may not matter (hence using an alias)18:27
loolNot link status, but interface up/down-ess for sure18:27
persiaRight, it only verifies the interface is up.  You don't need to connect a cable.18:28
loolIt will bring the interface up if it's taken up and vice-versa18:28
persiaThis is frustrating for many people because they want behaviour to be controlled by link status, rather than interface status, but in your use case it's convenient.18:28
loolBridges are cool when they work, but so ugly most of the time :-/18:28
persiaWell, I guess.  I think they are always ugly, and prefer to have internal routing on virtual interfaces, but that's even messier to configure.18:29
persia(as you need an internal router, etc.)18:29
loolOh well, I guess I'll go the full blown way, use libvirt, create a tap each time I start a VM, and route and NAT them18:32
=== asac_ is now known as asac
persialool: That's ideal, but annoying to set up.18:41
loolDefinitely :-(18:42
=== mawhalen_ is now known as mawhalen
GrueMasterdavidm: ping - I'm looking at RC3, and the psb driver versions are all over the place.19:32
GrueMasterdri modules are 0016, xorg module is 0014, kernel module is 0010.19:34
GrueMasterdavidm: ping - I'm looking at RC3, and the psb driver versions are all over the place.21:22
GrueMasterdri modules are 0016, xorg module is 0014, kernel module is 0010.21:22
GrueMasterwhat happened?21:22
davidmIt is not any different then RC-221:25
GrueMasterUnfortunately, RC2 slipped in under the radar.  I didn't know about it or RC3 until this morning's meeting.21:26
GrueMasterBut RC1 had all 0014 builds of the psb driver set.  That's why my concern.21:26
davidmGrueMaster, the next kernel update release will have the latest drivers, we had to go out this way to get into the hardy upgrade tree which is a huge win long term for updates and upgrades.21:48
davidmUnless you find some critical issue we are going to go final on this and go with the upgrade later, as I told Don last night.21:49
davidmAs far as I am concerned it's a release. 21:49
GrueMasteramitk:  I'm downloading the -19 kernel now.  Do you know what version of the psb kernel module is in this?22:42
amitkGrueMaster: 001623:31
GrueMasterI see that now, thanks.  Looks ok so far.  I sent you an email.23:32

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