[15:49] <seb128> hey there
[15:49] <seb128> where would I find details about mir not starting in an unity8 session?
[15:49] <seb128> .cache/upstart/unity8.log states "ERROR: QMirServer - Mir failed to start"
[15:50] <seb128> but no detail
[16:06] <olli> camako, ^
[16:06] <seb128> olli, thanks ;-)
[16:07] <olli> also got a question... is there a mini pc system you guys would recommend that works oob w/ Mir
[16:36] <seb128> log on http://paste.ubuntu.com/11763124/
[16:38] <kdub> olli, not sure about the system... if its supported by mesa well and has been around a while, i'd say that increases the chances
[16:39] <kdub> seb128, it does look like a problem with input_stub.so... don't know if that's changed recently, or if its an installation thing
[16:40] <seb128> kdub, that line is only a warning though?
[16:41] <seb128> the error looks like https://bugs.launchpad.net/ubuntu/+source/mir/+bug/1458689
[16:43] <kdub> hmm, maybe whatever fixed it isn't in overlay... (not familiar with the problem)
[16:43] <kdub> anpok, any ideas?
[16:44] <seb128> kdub, I'm using wily
[16:44] <alan_g> That looks like a problem I've seen when "server-platform" has drivers from Mir 0.12 while the executable is from Mir 0.14
[16:44] <seb128> so I guess it's just > vivid
[16:47] <seb128> that seems a redherring
[16:47] <seb128> I've moved input-stub.so away, same error abotu mir failed to start
[17:41] <seb128> olli, I'm a bit stuck on snappy personnal/unity8 and that mir not starting, if you can suggest/find somebody to help debugging that would be useful ;-)
[17:41] <seb128> it's probably a topic for tomorrow rather than today now though
[17:41] <olli> seb128, it's kind of getting late over therr
[17:41] <olli> there
[17:42] <seb128> olli, yeah, as said, a tomorrow topic
[17:43] <olli> camako, kdub, alan_g|EOD, dandrader, greyback all should be in a position to help
[17:43] <seb128> I can try in the morning, maybe some of the .au guys are still around and have a debug idea then
[17:43] <seb128> k
[17:43] <seb128> let's see if I've more luck tomorrow then
[17:43] <seb128> olli, thanks
[17:43] <olli> seb128, sorry, no idea why this channel went dark
[17:44] <seb128> likely people busy and not watching IRC ;-)
[17:44] <camako> seb128, Sorry I thought discussion was ongoing
[17:44] <camako> I see it's stalled..
[17:46] <dandrader> not sure I can help. I know pretty much nothing about snappy
[17:47] <seb128> dandrader, it's not really a snappy issue
[17:47] <kdub> has usc started?
[17:47] <seb128> dandrader, it's just that unity8 fails to start with the pastebin I gave earlier
[17:47] <camako> seb128, try to catch anpok tomorrow. He recently fixed it in the devel branch
[17:47] <seb128> I can start the demo shell from ssh and use gtk apps in it
[17:47] <dandrader> seb128, think I missed the pastebin url
[17:48] <seb128> unsure if I can start unity8 manually and how
 log on http://paste.ubuntu.com/11763124/
[17:48] <seb128> dandrader, ^
[17:48] <dandrader> "undefined symbol: _ZN3mir6events10make_eventEll17MirKeyboardActionjij"
[17:48] <seb128> yeah, I moved that .so away
[17:48] <seb128> same issue
[17:49] <seb128> and the demo shell is starting
[17:49] <seb128> so I think it's not blocking the server to start
[17:49] <seb128> that's wily btw
[17:49] <seb128> I should maybe try if unity8 works on a normal deb system atm or if mir/unity8 are just busted on desktop for everyone
[17:51] <dandrader> hmm... it's been a while since I last tried running unity8 on the desktop. Gerry (greyback) would be the best person to ask, as I think he has done a lot of "unity8 on desktop" work recently but he's on holidays this week.
[17:53] <seb128> k
[17:54] <kdub> also, if you look at the --help for a demo server
[17:54] <kdub> any of those debug options can be prepended with MIR_SERVER_<option>=<value> in the environment variable
[17:54] <kdub> to get more info
[17:55] <seb128> the demo server work
[17:55] <seb128> I can start it fine and start a gtk application
[17:55] <seb128> and use it
[17:55] <seb128> it's unity8 which fails
[17:55] <seb128> and I'm unsure how to debug that
[17:55] <seb128> starting it from a vt like the server doesn't work
[17:56] <seb128> and starting it as a client with the demo server socket exported either
[18:01] <kdub> well, something like MIR_SERVER_DEBUG=true might give more clues
[18:11] <anpok> oh input-stub not linking to mir client
[18:11] <seb128> https://bugs.launchpad.net/ubuntu/+source/mir/+bug/1458689 suggests it has a fix
[18:11] <anpok> could input-stub be outdated?
[18:11] <seb128> but maybe that didn't land in wily yet
[18:11] <anpok> ah right alan fixed that
[18:12] <anpok> yeah time for a release
[18:17] <anpok> seb128: on which hardware do you try to start?
[18:17] <seb128> anpok, dell inspiron 11 3000
[18:17] <seb128> but that's my test machine for unity8 for some cycles, works fine usually
[18:17] <seb128> and the mir demo server works fine as well atm
[18:18] <seb128> so I think it's not a driver/mir compat issue
[18:18] <seb128> I'm just trying to get the new snappy personal image working
[18:18] <seb128> it could be an issue in the image or some missing permissions/files, but not sure how to debug it
[18:18] <anpok> but you dont run it through kvm?
[18:19] <anpok> oh ok you do
[18:21] <seb128> no, I don't, it's a real machine booting on an usb key
[18:26] <olli> how good or bad is Intel Iris 6100 supported atm?
[18:31] <anpok> hm
[18:35] <anpok> seb128: do you see anything interesting in ~/.cache/upstart/unity8.log*?
[18:36] <anpok> I just tried to launch it non-nested: and I got what():  Nested Mir Display Error: Failed to update EGL surface: EGL_BAD_DISPLAY (0x3008)
[18:36] <anpok> which should not happen
[18:37] <anpok> is qtmir somehow requiring to be stared nested?
[18:39] <anpok> seb128: do debug the startup - you could try what /usr/bin/ligthdm-unity8-session does .. but for that you first need to start some mir server.. mir_demo_server ... or usc (with --from-dm 0 --to-dm 1) and --file something ..
[18:39] <anpok> s/do/to
[19:08] <seb128> anpok, http://paste.ubuntu.com/11763124/ is the unity8.log
[19:09] <anpok> oh
[19:09] <seb128> starting the mir server and then a gtk application works
[19:09] <seb128> unsure how to start unity8 saying to use the mir server
[19:10] <seb128> unity8 is a mir compositor no?
[19:12] <anpok> yes a nested one, the shell script /usr/bin/lightdm-unity8-session shows which environment variables to set..+ you need to set MIR_SERVER_HOST_SOCKET and set it to the socket path specified as --file  param to the mir server.
[19:19] <anpok> seb128: or..
[19:21] <anpok> add the environment variables: MIR_SERVER_{DISPLAY_REPORT,COMPOSITOR_REPORT,CONNECTOR_REPORT,MSG_PROCESSOR_REPORT,SESSION_MEDIATOR_REPORT}=log to that ligthdm session script
[19:59] <bschaefer> hello, does qtmir talk to USC, or does qtmir do the entire compositing thing?
[20:01] <bschaefer> as it seems to create a platform/mirserver (not sure if it ends up talking to USC or its just the actual nested server unity8 starts up)
[20:01] <anpok> the nested server talks to the mirserver in usc
[20:01] <bschaefer> anpok, cool thanks!
[20:02]  * bschaefer trying to understand the USC/nested server stack...
[20:02] <anpok> but the plan is of course to delay the composition as much as possible and offload as much as possible onto android overlays/hardware planes
[20:03] <anpok> but right now a nested server just emits a single buffer for its surface
[20:04] <bschaefer> yeah it looks pretty simple mir server wise
[20:04] <bschaefer> as the USC does a lllot more
[21:46] <racarr> AlbertA: Please do. Thanks :)
[21:46] <racarr> re papi pushing
[21:49] <AlbertA> racarr: thanks
[22:35] <bschaefer> robert_ancell, soo looking at making a barebone compositor ... leads me to wondering where/who does the actual setting of a session?
[22:35] <bschaefer> ie. i see asio handling it, but who sends it the messages?
[22:35] <robert_ancell> bschaefer, setting of the active session?
[22:35] <bschaefer> robert_ancell, right, as ... i see this "opening session mirdemo"
[22:35] <bschaefer> but i only 1 see frame
[22:36] <bschaefer> then it goes back to the spinner/black
[22:36] <bschaefer> robert_ancell, if im using USC
[22:36]  * robert_ancell looks at the u-s-c repo
[22:36] <bschaefer> robert_ancell, as i cant seem to get the session handler to allow the mirdemos to render (for more then 1 frame)
[22:37] <bschaefer> do i just need to send SESSION_CONNECTED? (I also dont see that message doing anything in lp:lightdm expect log it)
[22:37] <robert_ancell> hmm, I don't know how the WindowManager stuff works. That's all new since I last worked on USC.
[22:38] <bschaefer> i see, shoots, as it would be nice to get USC working in a standalone way
[22:38] <bschaefer> vs creating a barebone which shares a lot of the same code
[22:38] <robert_ancell> bschaefer, right, we're not waiting for the session to connect in LightDM, so you shouldn't need that (at the current time)
[22:38] <bschaefer> robert_ancell, the over difference i see is mir remove the default config that was once ued
[22:39] <bschaefer> robert_ancell, right, soo i hacked it so it ensures the spinner to hide
[22:39] <bschaefer> but it still only renders 1 frame then goes black
[22:39] <bschaefer> robert_ancell, cool, thanks ill have to dig around some more!
[22:39] <robert_ancell> bschaefer, if you pull in an old version of u-s-c it didn't have a spinner. So all that extra UI might be complicating things
[22:40] <bschaefer> robert_ancell, true, i should find a version that was after the mir/default_server_config was remove
[22:40]  * bschaefer tries
[22:50] <bschaefer> robert_ancell, shoot, the spinner was added back when  #include <mir/default_server_configuration.h>
[22:50] <bschaefer> still existed
[22:50]  * bschaefer tries a newer version
[23:33] <bschaefer> robert_ancell, soo looking at the USC code some more, it seems that the message set active session is sent from lightdm? (or the DM fd)
[23:33] <bschaefer> from_dm_pipe is what is hooked into for on_read_payload, which then it can receive events such as
[23:33] <bschaefer>  USCMessageID::set_next_session
[23:34] <bschaefer> robert_ancell, is that just done simply when reading the seat default config files?