/srv/irclogs.ubuntu.com/2012/03/21/#ubuntu-touch.txt

=== MacSlow is now known as MacSlow|lunch
=== dandrader is now known as dandrader|afk
=== dandrader|afk is now known as dandrader
cnddandrader, can you review my grail merge proposals soon?14:56
cndwe need to get them in, and your fix in today14:57
cndbefore the beta 2 freeze tomorrow14:57
dandradercnd, sure. Right after I finish the refactoring I'm doing on my merge proposal as per your comment14:57
dandradercomments14:57
cndcool14:57
cndthanks :)14:57
tvossSpending some quality time with Jenkins and Google Test/Google Mock, Prepared packaging for patched chromium15:15
bregmaI got utouch-geis up-to-speed on libxorg-gtest-dev, now back to the grind on #904731 and #944822, pending merge-request-reviews15:16
dandraderI'm refactoring the test in my merge proposal for utouch-grail according to comments received from Chase15:17
cndI'm fixing up utouch-frame for the new xorg-gtest, and then I'll be releasing frame and grail upstream and for precise15:18
cndSatoris won't be around for standups on wednesdays for a while15:18
cndjust fyi15:18
bregmaWednesdays have always been special to his people15:19
cndbregma, just reviewed your branch15:26
cndI'm stumped by this failure: http://paste.ubuntu.com/893791/15:53
cndit attempts to link a program15:53
cndit complains about not finding the evemu symbols in libxorg-gtest.a15:53
cndbut I have -lutouch-evemu before libxorg-gtest.a on the command line15:54
bregmalink order is significant15:54
cndyeah, and I think I have it in the right order15:54
bregmano, it's in the reverse order15:54
cndargh15:55
cndyeah15:55
cndI got switched up15:55
bregmanow fight with libtool to get things in the right order15:55
bregmaalthough that bug ni libtool was probably fixed years ago15:56
=== dandrader is now known as dandrader|lunch
cnddandrader|lunch, I just reviewed your branch16:29
cndlooking good :)16:29
cndalright, I've got all the merge proposals out that I wanted to get into beta 216:36
cndas soon as they are all merged into grail and frame I'll be making releases16:37
cndbiab16:37
=== dandrader|lunch is now known as dandrader
Satoriscnd: ping18:11
cndSatoris, pong18:11
cndoh right18:11
cndsorry!18:11
cndI should have replied18:11
dandradercnd, about https://code.launchpad.net/~chasedouglas/utouch-grail/fix-tap-touch-accept/+merge/9847818:21
dandraderwhat's the effect of accepting of rejecting a touch that has already ended?18:22
dandraders/of/or18:22
cnddandrader, all touches must be accepted or rejected no matter what18:32
cndif a touch has ended, then when you reject, the entire touch sequence will be replayed to further X clients18:32
cndso imagine you have a finger paint app open18:32
cndyou tap on the screen with one finger18:33
cndthe touch ends before grail rejects the tap18:33
cndgrail still needs to reject it, and then the X server will resend the touch event sequence to the finger painting app18:33
dandraderChickenCutlass,  you mean it will resend the touch event sequence to a X client other than the finger painting app, right?18:35
dandradercnd, ^18:35
dandraderdamn auto-completion18:36
cnddandrader, no, it will resent it to the finger painting app18:37
cndhere, I'm assuming grail is being used through unity18:37
cndso unity's grail touch grab is rejecting the touch18:37
cndand then the finger paint application will receive the touch sequence18:37
dandraderah, ok18:38
=== dandrader is now known as dandrader|afk
cnddandrader|afk, bregma: I need one last review: https://code.launchpad.net/~chasedouglas/utouch-frame/fix-visibility2/+merge/9872520:27
cndit's necessary before we make a release of utouch-frame20:27
=== dandrader|afk is now known as dandrader
cnddandrader, btw, you don't need to fully resubmit a merge proposal20:29
cndmerely updating the branch will suffice :)20:29
cndyou can signify that you've fixed things and it is ready for review again by leaving a comment with a review type of "Resubmit"20:30
dandraderok, that will probably make the review simpler as you can just look at the new commits containing the changes caused by the review comments20:33
dandraderthan I can squash them all before pushing to trunk20:33
dandraders/than/then20:33
cnddandrader, you don't really need to squash them even20:39
cndyou just merge them in20:39
cndand they'll appear to be one commit20:40
cndthen you can still drill down into the history if you really want20:40
dandraderbut those commits that are just fixes coming from review comment just clutter the history20:45
cnddandrader, if you don't want to see the clutter, then just diff from the beginning to the end of the rview20:51
cndwhich will normally be the diff of the merge itself20:51
cnddandrader, your testcase is failing20:51
cndit's in the jenkins instance20:51
cnddandrader, I can confirm the failure here20:56
cndI need to grab a sandwich20:56
cndthen I'll try to figure out what's going wrong20:56
cndit looks like bad events are being fed into frame though...20:56
cndthat would cause the "Warning: failed to get previous touch value"20:56
dandraderdamn, I'm not getting any failure here with current trunk versions of utouch-frame and utouch-grail...21:12
cnddandrader, I don't have the current frame installed, let me double check21:13
dandraderfrom the error it seems that nothing was played21:15
dandraderas the slice checker was still on its first state, which check for a Begin slice from the touch gesture21:16
dandradercnd, have you run that ParallelAtomicGestures test before on your machine?21:17
cnddandrader, not until just now21:17
cndI assumed it would work21:17
dandrader:)21:17
* cnd can't wait until we have proper jenkins support21:18
dandradercould you check if the recording plays correctly on your machine?21:18
cndyeah, I'm just setting up frame right now21:18
cndfrom scratch21:18
cndto be sure21:18
cnddandrader, with the latest utouch-frame it fails in the same way here21:19
cnddandrader, I just noticed that you copied in the asynchronous playing of the recording21:23
cndwhich I just removed from all the other tests :)21:23
cndI should have caught that in review21:23
cndbut sadly it's not the cause of the failure21:24
cnddandrader, here's my output with grail debugging: http://paste.ubuntu.com/894298/21:24
cndlet me know if you can make anything out21:25
* dandrader reads...21:25
dandraderthat touch 4 is coming way too soon21:28
dandraderwith the same timestamp as touch 321:28
dandradercausing all gestures to be canceled before they have a chance to do anything21:29
dandraderwhen I did the recording the 4th touch should come just by the end of the recording21:29
* cnd tries to read the recording21:30
cndthe fourth touch starts at 1331926701.18059621:32
cndthe beginning is 1331926700.83565721:32
cndthe end is 1331926701.54899121:32
cndso it's about half way between the beginning and the end21:32
cndbut the whole recording is less than one second long21:33
cnddoes that seem right?21:33
cndin the debug output, gestures 0, 1, and 2 start at time 69138945021:34
cndthe first touch began at time 69138945021:34
cndsame time21:34
cndthe fourth touch begins at time 69138945021:35
cndwhich is the same time again...21:35
cndthat doesn't seem right21:35
dandradercould you try to play it with evemu-play and watch it with mtview?21:38
cndyeah,21:40
dandradercnd, http://ubuntuone.com/3zGHZA1ILFUm6LgFrhEYOZ21:41
dandraderthis is what I get in the end21:41
dandraderthe pink blot is the fourth touch that comes late21:41
cnddandrader, using evtest, I see the three touches begin at time 1332366154.522432, and then the fourth touch begins at time 1332366154.879781,21:44
cndso it is coming out of evdev about .35 seconds later than the first three touches21:44
bregmaarthritis slows you down21:45
cndwhich is accurate, according to the recording21:45
cndso the touch events are getting coallesced in the x servre21:45
cndserver*21:45
cnd0.35 s is a long time for the server to be sitting around...21:46
dandradersince grail now deals with timestamps, it would be nice if in the tests we could feed the events directly to grail instead of sending via an evdevice, through xserver, in realtime. Then all tests could be run in a split second and there would be less external interferences in the tests (if any)21:51
dandraderI wonder if we are running the same version of xserver...21:52
cnddandrader, then we would be wondering why gesture recognition isn't occurring properly in real life :(21:53
cndwe could do what you suggest, but we would still need integration tests21:53
cndI'm looking at the X code21:53
cndand it gets its time from either clock_gettime() or from gettimeofday()21:54
cndand events are processed in sigio context, so they should be getting their timestamps immediately21:54
cndso what's going on...21:54
* cnd goes to build a debug server with timestamp printouts21:54
dandraderI think it would be good to have both kinds of tests.21:58
cndyeah, no doubt21:59
cndbut I think if we have only one, they should be integration21:59
dandraderI agree21:59
cndwe need to find the resources for more testing :)21:59
cndactually, something doesn't seem right...22:01
cndwait, nm22:02
cndI'm wondering if the first <X> events aren't being posted22:03
cndthey are just getting dropped until the fourth touch comes along22:03
cnddandrader, is this on a trackpad?22:03
cndI know what's going on...22:03
dandraderyes, the apple wireless one22:03
cnddandrader, the default X synaptics properties are filtering touch events until 4 touches are present22:04
dandraderah! one thing is that I don't have the synaptics x driver installed on my machine!22:04
cndaha22:04
cnddandrader, if you redo the recording with 4 and 5 touches instead of 3 and 4, it should pass the test just fine22:05
cndthough this raises the question of why in ubuntu X synaptics is withholding 3 touch gestures22:06
cndit shouldn't be doing that22:06
cndhmm, I'm getting touch events22:09
cndoh wait22:10
cndno, I'm not22:11
cndwhy...22:11
cndok, there's a bad default in xf86-input-synaptics22:16
cndaha22:18
cndok, I know what has happened22:18
cndI need to upload a fix to xf86-input-synaptics22:18
cnddandrader, when I fix xserver-xorg-input-synaptics in an upload I'm making today, it should start working again22:19
dandradergreat!22:19
cndno need to make any changes, though the test needs to be made non-async22:19
cndso we can run it through valgrind22:19
cndif you're at your EOD, it can wait22:19
cndphew, I was worried X timestamps were so unreliable they were coallescing events .35 seconds apart :)22:20
cnddandrader, make sure you have synaptics installed from now on :)22:21
dandraderjust installed it and rebooted22:21
dandraderbecause I can't seem to be able to connect to my trackpad still22:21
cndhmm? you can connect when using evdev but not synaptics?22:22
dandraderbefore I installed the synaptics driver I was able to pair with it effortlessly. but after I installed I cannot pair. might just be a coincidence...22:23
dandraderwell, will get back to it tomorrow. but hey, in the end you found a new bug! :)22:24
cndbregma, will you be uploading utouch-geis soon?23:03
cndwe need to get the tap fixes into precise for beta 2, at the very least23:03

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