=== MacSlow is now known as MacSlow|lunch === MacSlow|lunch is now known as MacSlow [14:02] Satoris, the bug report looks nice :) [14:02] unfortunately, I'm getting some duplicate entries [14:03] you have them in your foo.html too [14:03] look at bug 963525 [14:03] when I run your script it has three entries for each task (dulwich and bzr-git) [14:04] oh wait [14:04] there's something buggy with the colunn sorter [14:04] ahh, nm, I didn't realize something [14:04] there's three tasks for that bug [14:04] for each task, there are three entries where there should be only one [14:13] * bregma drains his coffee cup [14:14] I think I have fixed all the "bugs" with touchscreens in X right now, but there is a design issue that prevents menus from working properly [14:14] working on utouch-frame generic backend, prototype of x-backend rewrite is working, iterating chromium patch, jenkins work (production and staging) [14:14] I'm trying to figure out if we should just say they are broken [14:14] and move on [14:15] or if we should change the XI 2.2 protocol slightly when explicit device grabs are used [14:15] cnd, what is the design issue? Just curious [14:15] when a touch begins, a list of listeners is created [14:15] touches and pointer events derived from them are only ever sent to that list of listeners [14:16] Bug report and bugfixing secret project. [14:16] however, the way menus work is it receives a button press [14:16] ah, and a menu opens up after the initial list of listeners is created? [14:16] then it creates a new menu window [14:16] and then it grabs the device [14:16] and it expects to receive events through that grab [14:16] cnd: you can't have only one line for each entry, because they can all have different status/importance/milestone/etc. [14:16] which it does for subsequent touches, but not for the touch that initiated it [14:17] Satoris, there should be one line per task [14:17] Or putting them in a tree structure, which is unsortable as discussed earlier. [14:17] I'm seeing 3 lines per task [14:17] but only for a few bugs [14:17] most are correct [14:17] dandrader, standups! [14:18] Hmm. That is strange. [14:18] I'm now debugging the patch that makes unity use geisv2 api with accept() reject() (regular recognizer) [14:18] bregma, if you're going to drink coffee by the water cooler, you need to chat too :) [14:18] I got my gesture-config stuff almost done yesterday then realized the config is per-subscription, not global, so now I have a whole lot more left to impement [14:18] dandrader, where do we stand with the geis touch coordinates? [14:18] test cases these days are the bulk of the work [14:19] I put bugs in an array. Probably need to change that to a dict so that duplicates go away. [14:19] Satoris, ahh, that sounds like a good idea :) [14:19] cnd, didn't start working on it yet. will do after I finish with the unity geisv2 update (ie. once it's working again) [14:19] dandrader, how long until you have geisv2 done? [14:21] took much more work than I anticipated. I wanted to expose that event->groupset->group->gestures hierarchy in GeisAdapter [14:21] dandrader, if it's not much more then that's fine, but we need to focus on fixing bugs for the release [14:22] jeepers cripes I just went to do an apt-get upgrade and it's 2 GB of downloads... has every single package in the distro been updated in the last 3 or 4 days? [14:22] cnd: is there anything else to add to the output? [14:23] Satoris, assignee would be good [14:23] Ok, will add. [14:23] otherwise, I think it's good :) [14:24] for a future version, I'm starting to think we want to separate bugs that are relevant to us, and bugs that are our responsibility [14:24] so the bzr-git signed tag support is relevant to us, but not under our control [14:24] we'll have to come up with a way to separate them though [14:25] but that's after we get the initial report up [14:29] I'll be biab [14:36] cnd: I'll leave now but Skype later? === dandrader is now known as dandrader|afk === dandrader|afk is now known as dandrader [15:45] cnd, https://code.launchpad.net/~dandrader/utouch-geis/lp984069/+merge/102337 [15:46] should I include a test as well? [15:55] dandrader, fix looks good, but yes it does need a test [15:55] dandrader, you can use one of the existing touchscreen device recordings in the geis testsuite [15:57] cnd, could you test it out (as explained in the proposal)? [15:57] dandrader, yeah [15:58] dandrader, I won't get to it for a bit thoug [15:58] meetings and such [15:58] ok === dandrader is now known as dandrader|lunch [16:55] bregma, wrt dandrader|lunch's branch, did you build and run the unity branch too? [16:55] if you were able to perform three touch gestures, then it should be all fixed [17:01] no, I have not rebuilt unity [17:01] I do want to know why having to have all three touches in the window is a requirement [17:02] it seems we discussed that and I recall the conclusion was that it's OK for one touch to be outside the window (fat fingers and small windows) [17:02] that's why Ikbel asked the question on the big === dandrader|lunch is now known as dandrader [17:16] bregma, I get several failures when running geis tests. do they all work for you? [17:19] you mean "make check"? [17:20] yes [17:20] yes, works for me on all my machines, at least on a first pass [17:20] well, if I leave my computer alone while the display is blank I get only one failure (a seg fault) [17:20] on subsequent runs I sometimes get the spurious failure [17:22] so the tests automatically switch to a tty and should be kept there? [17:22] I don't run them in a tty [17:23] the gtest-based tests shouldn't need one, the check-based tests don;t need one [17:23] you _do_ need to be root to run the gtest-based tests [17:23] me neither, but when I run them they seem to do so. My screen gets blank and I have to do Ctrl+Alt+F7 to get back to my desktop [17:24] yes, I'm running them as root [17:24] perhaps one of the captive evemu recordings are doing something bad to your desktop [17:27] hmm... now I get more failures but no tty swithc [17:27] that switch might be related to the fact that I'm running compiz from a tty [17:28] would be nice if those recordings could be played in-process through a fake utouch-frame layer [17:35] or something like that [17:59] dandrader, yes and no, I like having the recordings play through X because then I can double check that I haven't screwed anything up in the display server :) [17:59] and the necessary VT switch should be fixed in a future xorg-gtest release [17:59] if you have xorg-gtest build and installed from upstream git, then log out and back in, you shouldn't need a VT switch anymore [18:39] something is really wrong. tests in all utouch projects are failing for me. no utouch-grail test gets any event (such as device added) for instance [18:39] I wonder what has changed since the last time I've run those tests... [18:40] (successfully, that is) [19:10] dandrader, do you have a stale X server process maybe? [19:10] sometimes the test dies unexpectedly and the X server is left running [19:10] hmmm... let me check [19:10] that could cause an issue [19:12] cnd, I did! [19:12] good. now I'm back to a new set of failures and that odd VT switch :) [19:12] dandrader, hopefully that is resolved with an upstream X server patch and an update to xorg-gtest [19:12] I'm just waiting on keith packard to merge the patch into xserver [19:13] dandrader, I'm checking out geis and unity now :) [19:13] my frustration level decreased quite a bit now :) [19:14] (with your tip about the stale X server) [19:14] good, though I'm sorry you had to go through it :( [19:15] ah, yeah, that. forgot to mention it sometimes happens [19:15] cnd, btw, I think you're the appropriate person to reply to this one https://bugs.launchpad.net/unity/+bug/978378/comments/1 [19:16] ahh, I hadn't seen that comment yet [19:16] I'll respond [19:16] dandrader, btw, Ikbel was a previous member of the utouch team [19:16] he wrote ginn [19:16] just fyi [19:16] hmmm, is he still at Canonical? [19:17] that's the comment on the bug I mentioned above (see scrollback) [19:17] dandrader, no, he's studying full time [19:18] ah, ok [19:23] dandrader, hmm... the coords still seem wrong [19:23] I'll debug a bit [19:24] :( [19:25] oh wait [19:25] I checked out the wrong geis branch :) [19:28] dandrader, the geis change fixed it :) [19:28] \o/ [19:29] finally! [19:29] when straddling windows it doesn't work any more === dandrader is now known as dandrader|afk === dandrader|afk is now known as dandrader