bschaeferRAOF, o looks like cookie factory inst installing correctly?00:40
RAOFHah! Indeed.00:41
bschaefermore protobuf issues for amd6400:41
bschaefersad face until that gets fixes00:41
RAOFWe haven't added a libmircookie package to debian/control :)00:41
bschaeferRAOF, o duh, i added the nettle-dev to control00:41
AlbertAbschaefer: more? like what?00:42
AlbertAbschaefer: it's compiling and passing here with GCC (lp:mir)00:42
bschaeferAlbertA, more?00:42
bschaeferAlbertA, umm in jenkins00:42
bschaeferlp:mir compiles fine for me locally00:42
bschaeferthough its failing in clang00:42
* bschaefer should try that00:43
RAOFYeah, clang is known-broken with gcc-5 built stuff.00:44
bschaeferwell at lease its not my fault haha00:44
bschaeferRAOF, so we need to create a cookie factory package?00:45
bschaeferie library?00:45
RAOF+ libmircookie-dev00:45
* bschaefer attempts00:46
* RAOF originally had this as an external project called libspoofless, but that was just soooooo much annoying overhead.00:46
bschaeferRAOF, might need to be worded a bit better :)00:54
RAOFI can probably update the description in the branch if you like :)00:55
bschaeferRAOF, pushed if you want to change it00:55
bschaeferhaha yeah00:55
bschaeferinstall files there00:55
=== chihchun_afk is now known as chihchun
MirvRAOF: are you still about? can I get a core dev ack on Mir https://ci-train.ubuntu.com/job/ubuntu-landing-028-2-publish/lastSuccessfulBuild/artifact/mir_packaging_changes.diff ? (the debian/ files, the rest are just for information)05:16
Mirvseems ok changes and cleanups for me05:16
RAOFMirv: Yeah, seems sensible to me.05:33
MirvRAOF: thanks! I still need archive admin preNEW review though to ack the adding of binary packages, but 1/2 completed.05:34
duflutvoss: Hey I found Jolla are to blame for our smoothness issues, kind of :)  https://bugs.launchpad.net/qtmir/+bug/148634105:56
ubot5Ubuntu bug 1486341 in unity8 (Ubuntu) "Touch scrolling is stuttery under Unity8 (but only while you're touching it)" [Undecided,New]05:56
RAOFOh, goody. That kernel panic corrupted my bzr working tree.07:14
Mirvduflu: it'd look to me that at least the MP linked to from QTBUG is only found in Qt 5.5 while the 5.4 branch has a crash fix that mentions it but apparently not the touch compression feature itself (looking at git logs at git://code.qt.io/qt/qtdeclarative.git)07:55
dufluMirv: Yeah the QTBUG might not be related. But the introduction of the feature occurred in 5.4 AFAIK07:56
dufluI know because Gerry told me about it early last year07:57
dufluAnd it was about to be released then07:57
Mirvyeah I also think I remember some earlier discussions, it's probably an earlier feature that has been later refined07:58
dufluMirv: Even if it worked properly, it would be redundant with what Mir already does07:59
RAOFNot quite; it can align event processing with Qt's renderloop, which Mir doesn't do.08:03
dufluRAOF: Nice in theory. In practice it's broken and everything works better without it :)08:05
greybackduflu: interesting findings on the touch scrolling, could it be the combination of the 2 input resampling engines being the problem, more than Qt's one alone?08:17
duflugreyback: Thought of that and apparently no. Turning off Mir's and just leaving Qt's the problem remains08:18
greybackduflu: ok08:18
duflugreyback: But please do try for yourself08:18
greybackwill do08:19
duflualf: OK, remind me who to ask to fix up the CI scripts?08:19
duflugreyback: Where in the architecture is it? Would apps have run though the compression twice or just once?08:24
greybackduflu: touch compression happens in each QQuickWindow. So the unty8 window will do compression before dispatching events to the qml scene.08:28
duflugreyback: I mean full stack. Just once then?08:28
greybackapps also have a QQuickWindow, and will use same compression logic08:28
greybackso no, multiple times08:28
dufluThat's great news. We might be able to go one better still08:29
alfduflu: it's usually fginther09:00
alfduflu: that fixes things for us09:00
duflualf: OK, thanks09:00
duflugreyback: Is there a way to use desktop_file_hint without having a valid file?09:01
greybackduflu: having a valid file is the point :)09:01
greybackjust point it to any existing file09:02
duflugreyback: Hmm, doesn't work. I might be using an incompatible Mir version or something09:02
greybackis unity8.log printing anything?09:02
greybackif client connection rejected, unity8 prints a REJECTED notice09:03
greybackjust use something like /usr/share/applications/gedit.desktop09:03
duflugreyback: I get the app starting animation indefinitely09:03
greybackduflu: something else wrong so, that means the client has connected, but not drawn its first frame09:04
duflugreyback: Arg. We might have broken protocol compatibility09:04
greybackI think we've a bug where actually a couple of frames need to be drawn09:04
duflualf: How do you feel about it landing (by hand) and then CI getting fixed?09:15
dufluIt would have to be in that order09:15
duflualf: Oh umm, or I could just hack the script to work09:16
dufluI might do that09:16
alfduflu: I wouldn't mind landing it first as long as CI gets fixed soon after (and CI is broken anyway so...)09:21
duflualf: Never mind. I have a solution that won't need CI changes now09:21
duflugreyback: Confirmed, we broke protocol compatibility in lp:mir (clients don't talk to existing servers any more)09:23
dufluNeed to bisect that09:23
greybackduflu: uh oh09:23
duflugreyback: Yeah, it was working at least a week or two ago09:24
alfduflu: seems like a good candidate for automated testing, something to consider for our new jenkins instance09:38
duflualf: Kind of the reverse of what we have in glmark2. That tests the previous release client against the newer server (I think?)09:45
alfduflu: only if there is an ABI bump, otherwise glmark2 just uses the update library of the same ABI09:50
dufluHow very predictable09:50
=== alan_g is now known as alan_g|lunch
=== vesar is now known as vesar_lunch
kdubalf, with lp:~afrantzis/mir/buffer-bind-to-render-image how does a mgm/mga::Buffer switch between gl binding and another sort of binding?11:27
=== vesar_lunch is now known as vesar
=== vesar is now known as vesar_
=== vesar_ is now known as vesar__
alfkdub: In a future MP, we will have a Platform::create_buffer_allocator(RendererType const& renderer_type), so the buffer allocator will know to create buffers supporting the renderer11:30
kduband, just guessing, will have something like: mgm/mga::Buffer::Buffer(..., std::shared_ptr<RenderTypeBinder> const&) ?11:32
alfkdub: Each platform can handle this as it sees fit internally, but what you propose is a reasonable approach if a platform supports multiple renderers. For the first step it's not needed since the existing platforms will just support "gles2".11:36
=== alan_g|lunch is now known as alan_g
=== tedg is now known as ted
=== dandrader_ is now known as dandrader
=== dandrader is now known as dandrader|afk
=== dandrader|afk is now known as dandrader
=== chihchun_afk is now known as chihchun
=== alan_g is now known as alan_g|EOD
kenvandinekgunn, i don't know how to get debs of mir built that i can test on my device, how would you guys feel about adding my branch to silo 0 to so i can test it with my touchpad?18:07
kenvandinesound only affect touchpads, if it doesn't work we can reconfigure it again :)18:07
kgunngreyback_: ^ ?18:07
kgunnkenvandine: in general i don't mind18:07
kgunni suppose we'd prefer to merge that into the silo0 branch for mir18:08
kenvandinesure, if it fixes it18:08
kenvandinebut i haven't actually been able to test it :)18:08
=== dandrader is now known as dandrader|afk
=== dandrader|afk is now known as dandrader

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