robert_ancellRAOF, MM8 seems to run better for some simple test cases with mir_demo_server00:13
robert_ancellstill falls apart if I plug a monitor while running unity or from the greeter though00:15
RAOFAh, good. Valgrind sees all!00:39
RAOFMir doesn't handle monitor unplug, apparently.00:40
RAOFUnplugging my HDMI monitor and then querying the display config crashes u-s-c00:41
* robert_ancell -> lunch00:56
robert_ancellRAOF, any thoughts on that unplug issue? Are you working towards a MM9?02:02
robert_ancellRAOF, Is there anything I can do to help here - multi-monitor isn't working on my system at all and I don't know of any specific bugs I can fix. If there's no new packages coming I02:28
robert_ancell'm not doing anything useful here02:28
RAOFrobert_ancell: I don't think there's anything to test that can't be more easily done by me locally at the moment.02:31
robert_ancellok, any idea when something will be testable?02:31
RAOFI don't think I can give a useful estimate.02:32
RAOFShould be soonish, but that depends on my understanding of the current problems.02:32
kgunnwhat' s up fella's i'm back02:45
kgunnanything to test ?02:45
robert_ancellkgunn, nothing at the moment02:47
* RAOF tries with the next set of debug spew.02:48
RAOFThat's *better*; now I merely need to find out where I'm setting the stride incorrectly.03:10
RAOFAnd that's where the pitch is wrong.03:15
dufluErm, both testing PPAs populated? Which one should I use?03:20
robert_ancellduflu, qa-testing is MM+bypass, qa-testing2 is just MM03:21
kgunnduflu: ....just to add, qa-testing2 is a non-starter atm03:25
* duflu cancels upgrade03:25
RAOFBut might soon start & win!03:27
kgunnRAOF: curious...why is qa-testing xorg mm8 & qa-testing2 mm7 ?03:27
RAOFkgunn: Because I uploaded mm8 to the wrong PPA03:28
dufluThis could cause some pinning confusion...03:28
kgunni love you guys....make me feel normal :)03:28
kgunnduflu: how so ?03:29
duflukgunn: My system was pinned for qa-testing. So switching to qa-testing2 might not get all the packages from it03:30
kgunnduflu: right...you'd have to purge first03:30
robert_ancellyep, always do a purge03:30
dufluHmm, the pointer lag that I previously attributed to software cursors is mostly gone when using a PPA with bypass enabled03:46
kgunnrobert_ancell: you updated qt-testing2 from archive basically ?...thanks for that03:46
duflu... mostly03:46
robert_ancellkgunn, no, just rebasing on archive when it changes03:46
kgunnrobert_ancell: yeah...what i meant...thanks for that03:47
kgunni was semi worrying about that03:47
kgunnbtw...did you see olli's ping earlier about u-s-c crashing on him ?03:48
RAOFWhen he removes a monitor?03:48
kgunnno it was just random use of xmir (i think)03:48
kgunnbut then apport said something like...can't report because03:48
kgunnversion of u-s-c changed ?03:49
robert_ancellkgunn, was he using a PPA?03:49
kgunn<olli> kgunn, tvoss|dinner_and_benchmarking seeing something weird03:49
kgunn<olli> while working on my kernel issue, I got an apport report of u-s-c crashing03:49
kgunn<olli> which in iteself sucks but is no surprise03:49
kgunn<olli> however I have #seat=unity03:49
kgunn<olli> and when proceeding to file the bug through apport, it complains that the bug can't be filed since u-s-c has changed since the crash occured03:49
robert_ancellmight be an old crash report03:50
RAOFOk. Lets see if mm9 + a new intel driver gets us all the way to spanned displays.03:50
dufluIs anyone testing non-intel? I can't with saucy :(03:53
RAOFIt almost certainly won't work with !intel at the moment; I'll get to that after intel works.03:53
dufluI can test radeon and nouveau on raring tho... Need to build everything by hand03:55
RAOFHm. Might be lunch time.03:58
dufluHm. Might be breakfast time04:03
robert_ancellHm. Might be dinner time04:04
olliHm. Might be bed time04:06
olliis there anything worth testing?04:07
tvossgood morning :)04:12
ollikgunn, I was bisecting kernels and it might have been just a weird side effect04:12
olli tvoss says: Hm. Might be breakfast time04:13
* duflu concurs. 04:13
dufluBrunch even04:13
tvossolli, indeed :)04:13
olliwell, it's beer o'clock somewhere04:13
kgunnolli: i thot you were in some remote forest or canyon or something04:14
ollikgunn, I will be in 12h04:15
olliright now though I am having ice cream and redwine04:16
olliwaiting for the miracle in #ubuntu-mir street04:16
olliduflu, anything worth giving a shot?04:17
kgunni think waiting on a mir-acle makes you gain weight...waffling between beer & choc chip cookies04:17
ollihah, kgunn just inspired me for the blog post once everything has landed04:18
* tvoss hands coffee to RAOF :)04:41
dufluolli: I know as much as you. RAOF and robert_ancell are in control of PPAs04:46
robert_ancellolli, there's no changes worth testing. RAOF is working on X changes and the PPAs are ready for that when available04:47
RAOFXserver MM9 should be in qa-testing2, but that's not going to work properly until xf86-video-intel mm504:48
RAOFWhich, assuming I've fixed the build properly this time, should be uploaded soon.04:49
* tvoss hugs RAOF 04:50
ollialrighty, will keep my fingers crossed04:50
olli(start getting calluses though)04:50
RAOFOk. Let's see if this works?05:01
kgunnRAOF: amd64 failed05:02
tvosskgunn, in qa-testing2?05:03
kgunntvoss: my bad05:04
kgunntoo many tabs05:04
tvossRAOF, did you upload mm5, yet?05:05
olliRAOF, qa-testing2?05:12
RAOFYeah, that's where it'll end up05:12
RAOFJust testing it locally.05:12
tvossduflu, did you see https://code.launchpad.net/~robertcarr/mir/weak-session-for-session-mediator/+merge/18192405:12
olliRAOF, ok05:12
RAOFJust worked out why it tends to turn my HDMI output off, too, which once fixed will make it easier to test :)05:13
duflutvoss: I intentionally don't look at code reviews until I have a longer stretch of hours to fuss over them. So... soon05:13
tvossduflu, ack :)05:13
RAOFqa-testing2, xorg-server MM10, xf86-video-intel MM505:28
RAOFYou'll need to wait for them to build, obviously.05:29
RAOFWell, not obviously. But you will need to wait for them to build :)05:29
dufluNow obvious05:29
kgunnin the words of tom petty "the waiting is the hardest part"05:30
* kgunn subscribes to the theory there's a song lyric for every situation05:31
dufluRAOF: You silently switched from qa-testing to qa-testing2 ... is that because bypass is still suspicious? or you want to keep it simple testing one thing at a time?05:31
RAOFOh, because bypass is still suspicious.05:31
duflukgunn: Generally true.05:31
dufluRAOF: Suspected of ... (something I can test for?)05:32
RAOFSuspected of failing to send buffers appropriately.05:32
RAOFNow that MM works I can dig deeper there.05:32
dufluRAOF: How so?05:32
robert_ancellI have to go, I'll try and hop on later tonight and check in05:33
kgunnor duflu could dig there...while you dig on ati :)05:33
kgunnrobert_ancell: o/05:33
RAOFduflu: Because, when XMir is driving multiple displays, unity-system-compositor only shows the first frame since it was VT switched to.05:33
RAOFrobert_ancell: Have fun wherever :)05:33
duflukgunn: Due to multiple kernel bugs, I cannot test bypass on radeon. Only alf could05:33
kgunnduflu: even after the 3.11 arrival ?05:34
duflukgunn: Because of 3.11. Booting 3.11 on raring broke too much, and booting saucy on my radeon machine fails to boot at all for unknown reasons05:35
tvossRAOF, did you upload mm5 for the intel driver, yet?05:36
duflukgunn: So I was going to call that a blocker till alf tested it for me05:36
dufluRAOF: I got exceptions from USC yesterday suggesting bypass might fail due to outputs not having CRTCs assigned at times. Is that meant to be possible?05:38
dufluIt's not possible to reproduce with plain Mir. Only saw it with XMir + MM setup05:38
dufluWhich reminds me. I need to beef up our exceptions if it happens again05:39
RAOFtvoss: Yeah, I've uploaded MM5 of -intel05:43
tvossRAOF, in qa-testing2?05:44
RAOF...but I don't see it in the PPA, so I'll check my email.05:44
tvossyup :)05:44
RAOFHah! Which says ‘you sent it to qa-testing, dufus’05:44
kgunntvoss: so i was playing with lttng....trying to follow the http://unity.ubuntu.com/mir/component_reports.html05:45
kgunnso when i cd into /tmp/ i don't mirsession05:45
RAOFWhy is everything terrible?05:46
kgunnor rather...if i follow those instructions...what is substituted for <sub dir> in ....babeltrace /tmp/mirsession/<trace-subdir>05:46
* RAOF cancels the amd64 xserver build that hung in the xvfb-run test05:46
tvossRAOF, kgunn moving over to the new house with a second monitor with me :) back in ~2005:49
RAOFHm. Damage is slightly off.05:58
RAOFBut mostly ok.05:58
RAOFi386 build, take 2, is done.06:01
RAOFamd64 build is nearly done.06:05
RAOFHappy testing, gentlemen.06:05
RAOFI'm off to get groceries.06:06
tvossback again06:11
tvossRAOF, dist-upgrading06:12
tvosskgunn, can you check your lttng version? please see the remark at the bottom of the page06:13
kgunntvoss: hmm, says 2.1.1...06:15
kgunnbut thanks for that...that pointer to the 2.2+ might be an issue on my android effort06:17
ollikgunn, which PPA did RAOF suggest06:18
kgunnolli: ^06:18
kgunnolli: don't forget to purge first (if you had been using qa-testing....not qa-testing2)06:19
kgunnand of course updaet your pinning pref file06:19
ollikgunn, on it06:20
tvosskgunn, did you try the ppa, yet?06:21
kgunntvoss: yep...just incessant punting to the greeter06:21
kgunni wanna double check all my packages first thos06:21
kgunnjust in case06:21
* olli finishes the install of qa-testing206:25
olliwhat do we expect to work?06:25
kgunntvoss: you ?06:25
kgunnolli: i would just try vga hotplug06:25
kgunnall my packages match :(06:27
ollikgunn, should I refrain from testing?06:27
ollior are you talking lttng?06:27
kgunnsecond data point can't hurt06:27
kgunnunfortunately....i'm talkin' bout qa-testing206:28
kgunnhmmmm....hdmi worked....after 1 punting to greeter06:29
kgunnall the sizes even look proper (mirrored)06:30
kgunnooo, it's pissed...did not like me unchecking mirrored06:31
kgunnjust hung06:31
ollican't find the vga cable06:33
olliam on dvi via dock06:33
olliso far it's just mirrored06:33
ollidoesn't detect 2nd display06:33
dufluAre we meant to having working hotplugging yet? Even when I had two monitors working this morning, I had to reboot before xrandr detected the second one06:36
* tvoss reboots06:36
kgunnhotplugging works on standalone X06:36
kgunnand at least that was the goal...to have parity06:36
duflukgunn: Yeah. But I mean have we implemented it yet?06:36
* duflu isn't sure06:37
kgunnat least mir mm (demo clients) was ok with this (i think...)06:37
* kgunn now doubts self06:37
tvoss_RAOF, are we supposed to have hotplugging working?06:39
kgunnok...so hotplug on vga (1 test) didn't work, hdmi hotplug punted me to greeter but then on login worked...06:39
kgunnfrom boot both vga & hdmi work in mirror mode06:39
kgunnooo, take that back hdmi worked in mirror mode, vga does not06:40
kgunnhdmi also worked when in mirror mode...deselect mirror + built in display off.....it went to 2nd monitor only06:41
kgunn(which is a lot like hotplug)06:41
kgunnand hdmi was ok, with switching mirror mode on (from the state of built in display off)06:41
kgunnok....that does it for me...i'm totally crashing (i feel asleep mid reboot)06:41
kgunnvga no good...hdmi seems pretty ok (mirror & going to second display only)06:42
kgunnhotplug wonky on hdmi, no good on vga06:42
dufluRAOF: Looks like XMir is honouring the desktop's alpha channel. I can "see through" the unity launcher when dragged06:42
olliwell, that didn't work so well06:44
olliI am off for today06:45
olligood luck gentlemen06:45
olliappreciate your work on the weekend06:45
dufluAlright. Later olli06:45
RAOFSee you.06:45
dufluHmm, X died and took me back to lightdm. But no logs show a crash... ?06:46
tvoss_duflu, RAOF hotplug with vga not working for me06:46
tvoss_I'm afk for some 30 minutes now, with you later06:47
RAOFduflu: What do you mean by "see through"?06:55
RAOFWell, HDMI hotplug worksish for me, although X doesn't seem to notice when its connected.06:55
dufluRAOF: Drag the launcher down and you can see a stationary one underneath. Perhaps an old frame or intended for a different screen06:55
dufluIt seems Compiz/Unity has some pixels with alpha < 1.0, which never used to matter but now they're translucent06:56
RAOFThey *shouldn't* be translucent. If they are it's a Mir bug, as Xmir chooses an xrgb pixel format.06:57
dufluHmm, well that's the least severe issue right now07:00
RAOFResolution changing seems to be working pretty well here.07:01
RAOFTurning the hdmi off, turning it on, changing the mode, moving it, all work.07:02
* duflu restarts with a more sane single output setup07:02
RAOFAlthough I *am* ussing the locally built packages07:02
* RAOF upgrades from the PPA07:02
dufluHmm, I don't get any resolution choices for LVDS. Only DisplayPort07:04
RAOFWe only pass on the modes the monitor supports.07:05
RAOFInternal panels almost uniformly lack scalers, so they only support their native mode.07:05
dufluReally? Every laptop I've tried had reasonable scaling for low-res modes07:08
RAOFNo, it's had faked scaling for low-res modes :)07:11
RAOFOr, rather, the scaling has been done on the GPU rather than by the panel.07:11
dufluRAOF: Yeah. For many people trying to run Unity on full-HD laptops, that feature is rather important07:17
tvoss_RAOF, duflu anything you want me to test?07:48
duflutvoss_: Can you confirm any of my bugs? https://bugs.launchpad.net/xmir/+bugs?field.tag=multimonitor07:49
tvoss_duflu, walking through them07:49
dufluIt's rather cool that you can still start native Mir clients inside unity-system-compositor and XMir just (correctly) gets out the way08:01
dufluOh wow, my translucent desktops are responding separately. It really is two launchers08:09
RAOFduflu: Can you modeset/change layouts with the xrandr tool?08:14
tvoss_duflu, you might want to give https://code.launchpad.net/~thomas-voss/glmark2/mir-no-deprecated-functions a spin08:14
dufluRAOF: I was just about to. Got side-tracked taking awesome blurry-cam video08:15
dufluRAOF: I suspect the blending issue is just "normal" for Compiz/Unity when you have two different displaybuffers physically overlapping08:17
RAOFLikely, yes.08:17
dufluHence, it's "pre-blended" before it hits Mir08:18
dufluRAOF: Yes, xrandr layout changes work.08:20
dufluOnly crashes with Control Center08:20
RAOFYeah. I suspect that it's actually broken with xrandr too, but we mostly don't hit it, whereas the little labels gnome-control-centre has makes it hit.08:21
dufluRAOF: Kinda. I tried some --off and then on again. And eventually broke it with xrandr08:23
dufluRAOF: Do the labels cause immediate damage or something strange?08:24
duflutvoss_: Cool. But I've never built glmark2 and might not get to it today08:26
tvoss_duflu, I will produce some numbers over the weekend for intel and amd08:26
duflutvoss_: Excellent. Can you see if a single build can support X and native Mir? Because glmark2 does. Then we can compare them directly. Even both running on the same USC instance :)08:27
tvoss_duflu, yup, you can build multiple flavors08:27
duflutvoss_: I mean build once, run anywhere08:27
tvoss_duflu, yup, that's what I mean, too :)08:27
duflutvoss: Nice. You could even run an X glmark2 and Mir glmark2 simultaneously on USC :)08:28
tvoss_duflu, exactly ;)08:28
tvoss_duflu, makes for nice blurry cam videos, too08:28
dufluThough they would fight for GPU08:28
tvoss_duflu, right, not so interesting for producing numbers08:29
tvoss_duflu, walked through the multi-monitor bugs and tagged them with affects me08:30
duflutvoss_, New --> Confirmed ?08:30
tvoss_duflu, nope, only affects me. Can set to confirmed08:31
tvoss_duflu, I will confirm against xmir08:32
tvoss_duflu, cannot check on displayport, though08:33
duflutvoss_: In theory the connector doesn't matter. Mir has no connector-specific code08:33
dufluAnyway, this is Intel, which means DisplayPort + DVI adapter = "HDMI" :)08:34
dufluThat's an old bug08:34
tvoss_duflu, ack08:40
* duflu wonders why we don't have an xrandr equivalent for Mir yet08:44
dufluWe have a demo client which shows it's possible. But that's just a toy08:45
robert_ancellRAOF, what's the state of MM10?08:59
duflurobert_ancell: Unchanged for a while. Bugs are against mir/xmir projects with "multimonitor" tag09:03
robert_ancellduflu, is MM10 considered mostly working?09:03
duflurobert_ancell: Hard to judge what mostly would be. I think maybe 30%09:04
duflurobert_ancell: The strangest one is X crashes if you use Control Center to change settings. xrandr works much better, but not 100%09:06
tvoss_RAOF, ping10:12
=== tvoss_ is now known as tvoss|test
=== tvoss|test is now known as tvoss_
RAOFtvoss_: Pong10:18
tvoss_olli, ping10:44
=== tvoss_ is now known as tvoss|test
ollitvoss|test, pong13:12
kgunnRAOF: vga connected at boot, just checked the xorg log its shows EE backtrace but nothing under it (like crashed before write out)14:36
kgunnright at "updated configuration" where it has output5/12 connected..then post-modeset config14:36
kgunnto be clear that was pre-boot vga connected....never got past greeter (greeter cycling)14:46
kgunnok...then i retested vga hotplug and got this https://pastebin.canonical.com/96336/15:06
kgunnseems different...like it almost reconfigured, but then went thru a bunch of keyboard/trackpad setup then bailed15:07
kgunnhuh...hotplug on hdmi worked (with adjusted res) but upon second connect/disconnect....its not alter the second screen res properly to fit it15:11
kgunnbut hdmi hotplug seems robust enough to have repeated connect/disconnect15:12
kgunnoh...and on hdmi hotplug...let it sit...not its frozen, can't even vtswitch15:14
kgunncorraction vga hotplug > https://pastebin.canonical.com/96337/15:25
kgunnseems hdmi hotplug will eventually fail in the same way15:25
rigvedhi everyone.17:37
rigvedi am running xmir on my 13.10 laptop.17:38
rigvedeverytime i try to open the dash or click on the shutdown menu item, my system freezes for a 2-3 minutes and then comes back to normal.17:39
rigvedeverything else seems to be running fine.17:39
rigvedcan someone help me diagnose the problem?17:40
=== Soupermanito is now known as GridCube_
robert_ancellRAOF, morning23:06
robert_ancellkgunn, around?23:23
kgunnrobert_ancell: hey kinda23:42
kgunnrobert_ancell: i was actually thinking about going to bed early and getting up around 4....since images don't usually show up until 1 -ish23:43
kgunnmy time...23:43
robert_ancellkgunn, I get slight improvements with MM10, looking at logs I don't think there's been any improvement until then23:43
robert_ancellChris said he'd be here in ~1:15, which means an image in ~2hrs perhaps?23:44
kgunnrobert_ancell: what are your results for hot plug?23:44
robert_ancellkgunn, I got it spanning on the greeter once, but it seems a bit random23:44
kgunnrobert_ancell: right...but you can't ever get past greeter?23:44
robert_ancellmost of the time it mirrors and u-g/XMir restarts quite a bit23:44
robert_ancellI can without the external monitor23:44
robert_ancellotherwise it's a bit random, sometimes it works, sometimes it doesn't23:45
robert_ancellI've only seen mirroring inside Unity23:45
kgunnrobert_ancell: right...i roughly see the same...23:46
kgunnaltho hdmi more stable than vga....23:46
kgunnwhich i never would've bet on23:46
robert_ancellI'm just testing VGA here23:46

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