kenvandineRAOF, hmm... i'm testing it on mako with silo 001:32
kenvandinenot mir_demo_server01:32
RAOFI don't know what input changes are in silo 0.02:01
=== chihchun_afk is now known as chihchun
bschaeferRAOF, would you know how trunk could possibly mess up reading umockdev recordings?02:34
bschaeferas ... once i merged from trunk the recordings no longer generate events in mir02:35
bschaefersoo either somehow trunk messed up new code you put in or the under the hood mir is different for handling events?02:35
=== chihchun is now known as chihchun_afk
RAOFbschaefer: I don't know. I'll pull your branch and investigate.02:51
bschaeferRAOF, alright, ill spend more time on it tomorrow if you've other things to get done02:52
* bschaefer slightly assumes he might have messed up fixing merge conflicts02:53
bschaeferbut i dont think so02:53
bschaeferRAOF, also i dont think we need the mir cookie for the basic surface validation meaning we can reduce a lot of code changing02:54
bschaefer(since a 0 should be a find value). Though i want to check if i even need to edit that make_event to begin with02:54
RAOFRight. You can certainly get part of the way through without actually plumbing in the real cookie.02:54
RAOFIt'll just be useless :)02:55
bschaeferyup, i was slightly lost when doing it at first and am trying to refine a lot of my changes02:55
bschaefermainly wanted to get all touch/keyboard/pointer to get cookies, but really only keyboard events + up/down events *need*02:56
RAOFWell, mostly.02:59
RAOFTouch begin/end probably need cookies, too. (As that's how “clicks” happen on touchscreens)02:59
bschaeferwell hmm right from what i can tell the validator does a very small touch event03:05
bschaeferonly 3 args03:05
* bschaefer needs to figure out the best way for that03:05
=== chihchun_afk is now known as chihchun
=== chihchun is now known as chihchun_afk
=== chihchun_afk is now known as chihchun
=== chihchun is now known as chihchun_afk
=== chihchun_afk is now known as chihchun
=== chihchun is now known as chihchun_afk
=== chihchun_afk is now known as chihchun
=== chihchun is now known as chihchun_afk
=== chihchun_afk is now known as chihchun
=== chihchun is now known as chihchun_afk
duflugreyback: On my way to EOD, but was wondering... Do you have any idea why scrolling with a constant finger touch in Qt feels like it's missing frames? Significantly less smooth than a fling...10:05
greybackduflu: it's something I've also observed, but thus far I've no good explanation10:05
dufluI would have said it's just keeping up with the finger by jumping. But then noticed Android doesn't have the problem and neither to mir-demos10:06
dufluneither do10:06
greybackI investigated if the input events received by the client were smooth - and they are10:06
duflugreyback: I know... we have some roughly equivalent mir-demos and they are smooth10:07
greybackduflu: yeah but I'm testing using unity8 here10:07
duflugreyback: On that note, does the --desktopfile trick still work?10:07
greybackduflu: yep10:07
dufluHmm, OK. Another day.10:08
greybackI want to improve qtmir's releasing buffer after swap, which might help10:08
duflugreyback: Yeah I've got it "done". Just held up by the related Mir fix waiting to land. Then I need to recombine and retest them10:09
greybackduflu: really? care to share?10:09
duflugreyback: Work in progress, and don't expect to see any benefit without the Mir fix coming in 0.16... https://code.launchpad.net/~vanvugt/qtmir/supersmooth10:10
greybackduflu: ok cool10:10
duflugreyback: Hmm, actually that looks old10:12
dufluI think I had a newer one10:12
greybackwell you're doing the right things10:13
duflugreyback: Oh, that is the latest10:13
ogra_duflu, fyi, i'm running my arale with GPU at full speed (using the sysfs hack from the mail thread) ... it ois a lot smoother and less stuttery ... but i'D guess it also eats about 10% more battery over time10:14
ogra_(i dont think we have any GPU scaling at all on the other phones)10:14
dufluogra_: Yeah I noticed raising the minimum frequency helped, but not as much as having the extra cores enabled10:14
ogra_we need ToyKeeper to measure the different scenarios for us ;) she has accurate power measurement tools10:15
greybackduflu: what mir commit would I need to test?10:18
duflugreyback: The current MP "earlier release"10:18
duflugreyback: https://code.launchpad.net/~vanvugt/mir/earlier-release/+merge/26732110:19
* duflu wanders off10:19
greybackduflu: thanks!10:19
alan_galf_: got headspace to discuss the remaining test failure I've got?11:35
alf_alan_g: sure11:35
alan_gI'm seeing an FD leak in some of the VariousDevices/EventHubDeviceEnumerationTest* tests. Specifically, those that call add_standard_device()11:36
alan_gThat allocates FDs in umockdev_testbed_load_ioctl() - which are never released11:37
alan_gBut none of this is new code - so how did it ever pass?11:37
alf_alan_g: Can you point me to a failure backtrace?11:38
=== chihchun_afk is now known as chihchun
alan_galf_: ^ any thoughts?11:47
alf_alan_g: Yes. I had put an exception for this case in detect_fd_leaks (checking for "g_dbus" in the backtrace), but in the new backtrace there is "bin doc src lib" instead of the g_dbus_* function names11:49
alf_alan_g: That's a thought, let me confirm this11:50
alan_galf_: that makes sense.11:50
alan_g(although the backtrace doesn't)11:51
alan_gI guess another exception for "mir_test_framework::UdevEnvironment::add_standard_device" would be specific enough11:53
=== alan_g is now known as alan_g|lunch
alf_alan_g: So, yes, I remembered that I was getting the g_dbus related errors locally, not in CI (probably because dbus is not running in the CI chroot?).12:07
alf_alan_g|lunch: ^^12:07
alan_g|lunchalf_: yep. I've MP'd the above (to see if anything remains that I haven't got locally)12:09
mterryalf_, hello!  Thanks for taking over that timeout branch for me.  I added a comment in your MP12:54
alf_mterry: No problem, thanks for the original fix :)12:55
alf_mterry: Let me try your idea...12:59
=== alan_g|lunch is now known as alan_g
alf_mterry: Your idea looks good. Would you like to incorporate the tests in your branch (+the max() idea)?13:11
mterryalf_, I'm in a meeting right now, so you can run with it13:12
alf_mterry: ok13:12
alf_mterry: Change applied. Thanks!13:22
alf_mterry: (and tested on arale)13:22
mterryalf_, sweet, I approved the branch from my side.  But maybe get a real mir folk to also look at it13:22
=== dandrader is now known as dandrader|afk
=== chihchun is now known as chihchun_afk
=== dandrader|afk is now known as dandrader
=== ogra_` is now known as ogra_
=== alan_g is now known as alan_g|EOD
=== dandrader is now known as dandrader|afk
=== dandrader|afk is now known as dandrader
=== ljp is now known as lpotter
=== dandrader is now known as dandrader|afk
=== dandrader|afk is now known as dandrader
bschaeferRAOF, for when you get on, thanks for fixing that test (had no idea i had to include that...)21:56
bschaefernow for the basic_suface validator21:56
bschaeferi would love to remove the need for the cookie factory, the only issue is the default event that is needed21:56
bschaeferif theres no prev events21:56
bschaefernow... i've clue when a default event would ever happen, or if its save enough to ignore the mac for a default event that we needed to generate21:57
bschaefer(im assuming yes its ok to ignore a mac for it since its just a bare bone vent)21:57
bschaefersoo now im starting to think the default touch event will have 0 pointer count meaning it will not have a down/up action21:59
bschaefermeaning we can remove the need to generate a valid mac for that21:59
* bschaefer double checks the default value for a make_event for a touch event21:59
bschaeferhmm  and now looking at that, the pointer_count variable for the new motion event is not being defined and everything is memset to 022:01
bschaefersoo yay no need for a cookie there since there will never be a up/down event or a default event if theres no prev touch event22:01
bschaeferRAOF, IGNORE MY PING22:01

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