/srv/irclogs.ubuntu.com/2013/10/21/#ubuntu-mir.txt

=== sam113101 is now known as sam113101_afk
=== sam113101_afk is now known as sam113101
=== sam113101 is now known as sam113101_afk
=== sam113101_afk is now known as sam113101
=== sam113101 is now known as sam113101_afk
=== sam113101_afk is now known as sam113101
sam113101hello03:46
tvossgood morning05:57
duflutvoss: Hello06:01
tvossduflu, hey there :) happy new week06:01
duflutvoss: Yes, happy Monday06:02
tvossduflu, how is it going?06:02
duflutvoss: OK I guess. You?06:04
tvossduflu, yup, all good :) 2nd coffee, so likely to be better after this cup :)06:11
dufludidrocks: Done. The new lp:mir has full history, and is separate to development-branch.07:36
didrocksduflu: ok, did you resync the commit id in the changelog?07:37
didrocksyou know the "last snapshot from rev xxx"07:37
didrockscan you change it that it matches the new commits if you have override the branch?07:37
dufludidrocks: The changelogs were already identical. Not sure what else you mean?07:37
didrocksoverriden*07:37
didrocksduflu: basically, you'll see a line with "Automatic snapshot from revision 1100"07:38
didrocksduflu: is that right? (I doubt it is if you changed the branch)07:38
didrocksif it's not, please just change that revision07:38
didrocksit's used to generate the automatic changelog07:38
dufludidrocks: Oh, I see. So the master copy is in the changelog head?07:38
dufluWill do07:38
didrocksthanks duflu ;)07:39
didrocksI guess it should be something around rev 113307:39
dufludidrocks: Propose it to dev-branch first for you to check?07:40
didrocksduflu: sure :)07:42
dufludidrocks: Or does the "Automatic snapshot" comment only make sense on lp:mir ?07:43
didrocksduflu: it makes sense for lp:mir, but if you want to be able to pull07:44
didrocksor you can exercise the other way07:44
didrockslike commiting that to lp:mir07:45
didrocksand then merging back07:45
didrocksas if it was a changelog bump07:45
dufludidrocks: development-branch should stay append-only for safety. It's lp:mir that pulls from development-branch. And development-branch merges from lp:mir07:45
dufluI'll do it in dev first...07:46
didrocksok07:46
dufludidrocks: Proposed. Now how do we tell Jenkins to pull rather than merge?08:09
didrocksduflu: that would be a question for fginther, mind emailing him?08:09
dufludidrocks: Mail sent. Mind reviewing? https://code.launchpad.net/~vanvugt/mir/begin-trusty/+merge/19194408:13
didrocksduflu: yeah, seems not the right one to me, let me look08:14
didrocksduflu: commented08:15
dufludidrocks: Oh I see. Change "Automatic snapshot from revision 1100" to 1133 without adding a new changelog entry?08:18
dufluIt would be accurate, but then inconsistent with the saucy branch08:19
didrocksduflu: don't worry about changelog accuracy, when merging from debian, some part of the changelog for most projects are removed (there are pros and cons)08:20
didrocksduflu: you can add a new changelog entry as well (to bump to 0.1.0), but in another MP maybe?08:20
dufludidrocks: I think it would make sense to not add new entries in development-branch. So just change the 0.0.15 entr?08:20
didrocksduflu: good to me as well, I have no strong opinion08:21
tvossduflu, did you test https://code.launchpad.net/~vanvugt/mir/fix-r1049-regressions/+merge/191776 on maguro and mako?08:24
duflutvoss: I ran out of time Friday night. I have mako, so will get to that. But reinstalling presently08:25
tvossduflu, ack, let me know if mako works, I can help out with maguro08:25
duflutvoss: The old version was well tested before 1049 too08:25
tvossduflu, ack08:26
dufludidrocks: MP updated. I'm wondering how we can get better accuracy in the changelog comments in future too. Seems like most bug fixes never get noticed/mentioned08:29
didrocksduflu: if they are autogenerated, the rules are simple:08:30
didrocks- if you change something in debian/changelog, the commit message isn't used08:30
didrocks- if you didn't touch debian/changelog, the commit message is used08:30
dufludidrocks: OK, so just tell the team to never touch changelog? If so then who/how does it get updated to 0.1.0?08:31
didrocksduflu: no, just ensure that if you modify it, you put all necessary informations08:33
dufludidrocks: So version bump entries come from development-branch and then leave it up to the robots on lp:mir?08:34
dufludidrocks: If this is wrong, I only have about an hour left to fix it... https://code.launchpad.net/~vanvugt/mir/begin-trusty/+merge/19194409:05
didrocksduflu: sorry, in meetings09:05
didrocksduflu: you didn't bump debian/changelog to 0.1.0?09:05
dufludidrocks: I was confused. I got the impression that the robots would figure it out. Or do those major bumps come from humans?09:06
didrocksduflu: no, major bumps are only handled by humans09:06
dufluOK09:06
didrocksso an additional entry UNRELEASED with 0.1.009:07
didrocksand fine to get it into :)09:07
dufludidrocks: But that entry does _not_ mention "Automatic snapshot..." ?09:07
didrocksduflu: right!09:08
dufludidrocks: So the robot(s) modify the head entry if it's "UNRELEASED"?09:09
penghuanHello, all, is there any doc about how to run the examples in mir? such as mir_demo_client_basic, etc. Thanks!09:11
didrocksduflu: exactly09:12
didrocksduflu: approved09:12
dufludidrocks: Thanks09:13
duflupenghuan: It's possible the docs are overcomplicated... but before you start, be aware "basic" does nothing. The other demos are more interesting.09:13
duflupenghuan: (1) start a server: sudo mir_demo_server_shell09:14
duflupenghuan: (2) Run a client: sudo mir_demo_client_eglplasma09:14
dufluBut you will need an ssh or other login to be able to do that :)09:15
penghuanduflu: OK,i'll have a try,thanks!09:15
penghuanduflu: When  run mir_demo_server_shell,i got an error "Address already in use", i'm running xmir ,it means i should stop xmir first?09:18
duflupenghuan: Yes that would be best09:19
penghuanduflu: Thanks!09:19
=== alan_g is now known as alan_g|lunch
=== dandrader is now known as dandrader|afk
=== dandrader|afk is now known as dandrader
=== alan_g|lunch is now known as alan_g
mhall119kgunn: where are the instructions for running Mir on 13.10?13:06
mhall119XMir and Unity 7 that is13:06
kgunnmhall119: lemme see if we've updated them13:07
alan_gkgunn: mhall119 this should be it: http://unity.ubuntu.com/mir/using_mir_on_pc.html13:07
kgunnmhall119: yeah...we did http://unity.ubuntu.com/mir/installing_prebuilt_on_pc.html13:07
mhall119thanks13:08
davmor2kgunn: daft question time?  What does xmir give us that is worth the effort and support for 5 years?  I'm just wondering if it might be time better spent perfecting mir ready for 14.1013:11
mhall119kgunn: where are the "If things go wrong, here's how to get back to plain Xorg"?13:11
tvossmhall119, you are after: http://unity.ubuntu.com/mir/using_mir_on_pc.html13:12
davmor2mhall119: remove the unity-* package that you install13:12
mhall119davmor2: one reason I can think of: We're going to need to support non-Unity DE's on a Mir system compositor for some time still13:12
mhall119davmor2: no simple config switch I can throw?13:13
mdeslaurdavmor2: I'd really like to be able to run apps and legacy apps that don't use gtk, qt, or sdl13:14
kgunnmhall119: the other method...is to comment out type=unity in the /etc/lightdm/lightdm.conf.d/10-unity-system-compositor.conf file (e.g. #type=unity)13:18
mhall119mdeslaur: I thought XMir was already able to run rootless X11 apps well enough, it was running a full X11 desktop and it's additional requirements that were causing problems13:18
mhall119thanks kgunn, that's what I was looking for13:18
kgunnmhall119: current xmir has no input support wrt mir (...its all still totally X)13:19
davmor2mdeslaur: let me see if I'm right though.  In 14.10 mir not xmir becomes the default to run unity and graphics in general right?  So xmir is only going to be used on 14.04 which then has to be supported for 5 years.  Unless xmir is going to be use for the foreseeable future then I would get it13:19
tvossmhall119, the main problem with XMir (specifically the X parts) is multi-monitor support, which is not required for a rootless scenario13:19
tvossdavmor2, we need xmir in 14.10 forward, too, to support legacy apps13:20
davmor2tvoss: okay now it makes more sense thanks13:20
mhall119kgunn: so how does that impact rootless X11 apps?  Does it mean they don't have input support atm?13:20
tvossmhall119, we don't support the rootless scenario right now13:21
kgunnmhall119: no, we don't have rootless yet13:21
kgunnwhat tvoss said :)13:22
kgunnmhall119: xmir is a truly full X session compositors sitting on top of mir only as a system compositor13:22
kgunnmhall119: when we do the work for rootless...the assumption is mir is both the session & system compositor13:23
mhall119ok13:23
kgunnso the rootless x is a windowed app into a full mir-only system13:23
mlankhorstsounds about right13:24
mhall119is there currently a way to run Unity 8 as a session in 13.10?13:24
mhall119on x86 laptop13:24
kgunnmhall119: https://unity.ubuntu.com/getinvolved/development/unity8/#running-unity13:26
kgunnmhall119: although....hang on...13:26
kgunnmhall119: this is better http://www.omgubuntu.co.uk/2013/08/unity-8-ubuntu-13-10-arrives13:28
mhall119kgunn: that will run it in a windows on my desktop, right?13:33
kgunnmhall119: yes, effectively13:34
kgunnmhall119: sans mir you realize13:34
mhall119I didn't13:52
mhall119but that's okay13:52
mhall119sorry, I know that when running Unity8 in a window on Unity 7 it won't use mir13:52
mhall119I was wondering if there was a way to run Unity 8 *as* the session on top of Mir13:52
mhall119so I can launch SDK apps inside of Unity 8 on my laptop13:53
kgunnmhall119: only on the phone....13:58
mhall119ok13:58
mhall119kgunn: are there any plans yet for getting an initial setup like that working?  Seems a necessary starting point for Unity 8 Desktop development to begin13:59
mhall119and yes,I know that Unity 8 Desktop *design* hasn't even started yet13:59
mhall119I'm just impatient13:59
kgunnmhall119: no problem...but i would question you....why not just launch your app within unity8 (on desktop w/o mir) ?14:00
mhall119can I do that?14:00
kgunnmhall119: i know its not completely real...but14:00
kgunnnothing will be real until we do full desktop14:01
mhall119last time I tried to open an app in the windowed Unity 8, it opened it in a Unity 7 window14:01
kgunnmhall119: when was that ?14:01
mhall119months ago14:01
=== alan_g is now known as alan_g|lunch
=== alan_g|lunch is now known as alan_g|tea
mhall119kgunn: okay, another question, has anybody looked at porting the SDL library to Mir?  There are a lot of games and apps that use it, and I believe it already supports multiple backends (X11, Windows, OSX)14:35
kgunnmhall119: yes...this is in progress (and actually working to a degree)14:36
kgunnmhall119: bschaefer has been doing this14:36
kgunnmhall119: the unfortunate thing is...some of the games pull in their own sdl bins14:37
kgunnmhall119: so you can't really plug-n-play like you'd think14:37
mdeslaurkgunn: are they statically linking to sdl, or they are shipping their own .so somewhere?14:39
kgunnmdeslaur: just went to read some of the correspondence on that...and you're right, i think we can load dynamically14:43
=== alan_g|tea is now known as alan_g
=== dandrader is now known as dandrader|lunch
=== dandrader|lunch is now known as dandrader
=== dandrader is now known as dandrader|afk
=== dandrader|afk is now known as dandrader
=== mdeslaur is now known as chuck_
=== chuck_ is now known as mdeslaur
=== alan_g is now known as alan_g|EOD
mhall119kgunn: so I followed the steps to run unity-system-compositor, but it isn't running17:09
mhall119does the intel i915 work on Mir?17:10
mhall119cat /var/log/lightdm/unity-system-compositor.log17:11
mhall119linkerlinker.c:1095| ERROR: Library '/system/lib/libGLESv2.so' not found17:11
sarnoldmhall119: tests/autopilot/unity_system_compositor/test_runtime.py  suggests yes17:11
mhall119I think maybe I have some problem with what's on my system...17:11
mhall119grep -i xmir /var/log/Xorg.0.log17:11
mhall119[ 63378.927] (WW) "xmir" is not to be loaded by default. Skipping.17:11
=== dandrader is now known as dandrader|afk
kdubanyone know what's going on with: https://jenkins.qa.ubuntu.com/job/mir-team-mir-development-branch-saucy-armhf-ci/17:43
=== dandrader|afk is now known as dandrader
racarrHowdy team17:58
racarrsorry about oversleep17:58
racarrHow is 14.04 going ;)18:00
kgunnracarr o/  mornin' ?18:02
kgunn:)18:02
kgunnmhall119: just guessing you don't have gles2 drivers on that system...which you would need for mir18:03
racarrMorning kgunn. Happy new release :D18:06
racarrI realized you are coming to California next week?18:06
kgunnracarr: yes18:06
kgunnracarr: altho...i've been in personal denial :)18:06
kgunntravel...ug18:06
racarryeah eww18:07
kgunni don't mind it...but i always lose so much sleep18:07
kgunnespecially sharing a room...18:07
racarrwell unless you manage to deny your way out of it18:08
racarrlets go out to dinner some night lol18:08
kgunnracarr: good idea! i'll let you know when i get out there what night will be more free18:08
kgunnbtw...seb & dednick will be out there18:08
kgunnif you just wanted to come have some f2f18:09
kgunnalong with saviq and others18:09
racarr:) sounds good18:09
racarryes, I think it would be good to come some time just to catch up with some people18:09
racarrbut don't know when it useful yet, so let me know.18:09
=== seb128_ is now known as seb128
mterryHeyo!  I'm messing around with nested Mir servers (i.e. through unity-system-compositor).  I see behavior where once the second nested server comes up, my first nested server aborts with "Bad file descriptor".  Is it possible they aren't sharing the host socket well?19:01
davmor2mterry: well don't mess with nested mir then :P19:03
mterry:(19:03
kgunnmterry: are you on arm/android or trying on desktop ?19:29
kgunnwrt nested mir19:29
mterrykgunn, arm/android19:29
kgunnmterry: there was a change around that....19:30
kgunnmterry: still digging19:33
kgunnmterry: there's this one...but not the one i was thinking... https://code.launchpad.net/~alan-griffiths/mir/socket-connection/+merge/18732619:37
mterrykgunn, ah hmm.  Yes.  robert_ancell was talking about adding a non-socket method for USC to talk to sessions (though unity8 and friends would continue to use a socket for their children)19:40
mterrykgunn, that branch looks like part of it19:40
mterrykgunn, maybe it's easier to use a new method than investigate why sockets don't work19:41
kgunnrobert_ancell should be on in a bit....and he might still like us just enough to give us some direction :)19:41
racarrlunch lunch19:42
racarrBack back.20:23
=== ara is now known as Guest23418
=== sam113101 is now known as sam113101_afk
mterrykgunn, robert_ancell approved my mir-fixes branch for u-s-c, so whenever you want to pull the switch on that abi bump, feel free to top-approve my branch21:32
kgunnmterry: rock on....thanks21:32
=== sam113101_afk is now known as sam113101
=== jono is now known as Guest78132
mterryrobert_ancell, this socket-connection branch that added the no-file mode to Mir...  It looks like it uses shared fd's, but USC child sessions aren't usually direct process children of USC, right?  And thus can't share the fd?22:35
robert_ancellmterry, lightdm gets the fd from u-s-c and leaves it open when forking off sessions22:37
mterryrobert_ancell, u-s-c can give an fd back up to its spawning parent?22:39
robert_ancellmterry, over the lightdm<>u-s-c protocol. It was a pipe but I changed it to a socket and it can now pass file descriptors across it22:39
mterryah..22:39
robert_ancellmterry, there's some clever dependencies required to upgrade these components synchronously22:40
robert_ancellmterry, all the work should be there, but when I last ran it XMir was failing to connect22:40
mterryWell... I'll give it a shot22:44
mterryrobert_ancell, is there any magic besides --no-file I need to provide?  Like, I assume that sets MIR_SOCKET to something magic that gets interpreted correctly without my effort?22:45
robert_ancellmterry, you set MIR_SOCKET to fd://n22:45
robert_ancelland then n just has to be the valid file descriptor22:45
robert_ancellall the details should be taken care of in libmirclient (assuming it works)22:46
mterryrobert_ancell, ideally lightdm would set MIR_SOCKET to the right thing.  I don't think I saw it doing anything fancy there last time I looked in trunk22:47
robert_ancellmterry, the branch changes it from setting MIR_SOCKET=/tmp/mir_socket (hard-coded) to MIR_SOCKET=fd://n22:48
robert_ancellmterry, what do you mean "that I need to provide?"22:48
robert_ancellbrb22:49
mterryrobert_ancell, I had missed that this was a branch we were talking about instead of trunk.  I see the work now in private-mir-connection.22:58
robert_ancellmterry, ah22:58
mterryrobert_ancell, it hasn't landed yet because xmir didn't work with it?22:58
robert_ancellmterry, correct22:58
robert_ancellmterry, also being too close to release for such a risky change22:59
mterryrobert_ancell, pfft, release is 6 months away!  ;)22:59
robert_ancelloh, it's fine for T!22:59
mterryrobert_ancell, well I'll play with it in my split greeter mode on the phone.  Maybe I'll discover something that explains xmir's problem23:00
robert_ancellmterry, nice, thanks!23:00
=== sam113101 is now known as sam113101_afk
=== sam113101_afk is now known as sam113101

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