qengho | RAOF: Hi, we're trading email about touchpads and XI2 touch events. IRC may be quicker. | 00:26 |
---|---|---|
RAOF | qengho: Sure. | 00:30 |
qengho | RAOF: I really want your nick to be ROUS, every time I read it. | 00:31 |
RAOF | qengho: If it makes you feel any better, it stands for Running Around On Fire. | 00:31 |
* RAOF 's cooperteam.net domain name may suggest the origin of that for the owner of sufficiently exotic trivia :) | 00:32 | |
qengho | That does comfort me. | 00:32 |
qengho | RAOF: I don't know the trivia, but sounds like some racing mishap. | 00:40 |
RAOF | qengho: So the problem is indeed that synaptics isn't giving you two-finger touches, and is instead returning scroll events? | 00:41 |
RAOF | ie: evdev worked? | 00:41 |
qengho | RAOF: Right. | 00:41 |
RAOF | So, this is where input becomes terrible. | 00:42 |
qengho | I can paste the events it *does* show, if you'd like. | 00:42 |
RAOF | For evdev, or synaptics (or, I guess, both :)) | 00:42 |
RAOF | Yeah, the events might be interesting. | 00:42 |
qengho | Er, I don't know what "synaptics" means. | 00:43 |
RAOF | Oh, the synaptics driver. | 00:43 |
RAOF | ie: before you deleted the “load synaptics for touchpads” xorg configuration. | 00:44 |
RAOF | But, in general - Peter Hutterer is the maintainer of synaptics, and would prefer very much to do gestures in-driver. Because you can't assume that clients will handle them, and sending both raw touches and synthesised scroll events becomes awkward. | 00:46 |
RAOF | I'm not _entirely_ sure that we can't do that sensibly, though. I'd like to try anyway :) | 00:47 |
qengho | RAOF: http://pastebin.ubuntu.com/7128187/ Unmodified xorg config, two-finger pinch. | 00:48 |
qengho | I'll have to go away and fabricate the other one. | 00:49 |
qengho | Modified to have no synaptics driver, I mean as other. | 00:49 |
RAOF | Right. | 00:49 |
qengho | Right, you need it? | 00:50 |
RAOF | I'd be interested to see it, but I could also get the same thing by charging some batteries, pairing my magic touchpad, and seeing whether it still panics my kernel. | 00:50 |
RAOF | And in that log you're seeing all the events on valuators 2 and 3, which are v. scroll and h. scroll respectively. | 00:51 |
qengho | Oh, I had that same problem, kernel panic. Thinkpad? | 00:51 |
RAOF | System76 | 00:52 |
RAOF | I think it's a bug in hid-magicmouse. | 00:52 |
qengho | Huh. I'm on a Darter. | 00:52 |
RAOF | Galago here. | 00:52 |
qengho | ...now. | 00:52 |
qengho | Back in 5min. | 00:53 |
qengho | ROUS: Well. Commenting everything out isn't right. Builtin trackpad doesn't do anything. | 00:57 |
* qengho restarts X. | 01:03 | |
qengho | RAOF: I'm realizing I don't really know what you wish me to change. | 01:09 |
RAOF | qengho: Oh. I wanted you to delete /usr/share/X11/xorg.conf.d/50-synaptics.conf and restart X; that'll switch your touchpad to using evdev, which definitely won't eat events to synthesise other events. | 01:10 |
qengho | RAOF: I didn't delete, but I added "#### " to the start of every line. That made neither builtin nor APPL bluetooth pad work. | 01:11 |
RAOF | Hm, that shouldn't happen. | 01:11 |
qengho | RAOF, right now, I added a new section that tries to match the device and assign it Driver evdev before the first one assigns synaptics. I don't think it's doing what I expected. Perhaps it's not matching. | 01:12 |
RAOF | Yeah, you'd need to do the matching *after* the synaptics one. | 01:12 |
RAOF | (/var/log/Xorg.0.log should show you which driver is getting loaded) | 01:13 |
qengho | I did it there too. I'm thorough. | 01:13 |
RAOF | I'm surprised that commenting out 50-synaptics.conf didn't work; evdev should still match in that case. Maybe could you throw me your Xorg.0.log at some point? | 01:14 |
qengho | Comment all, get log? | 01:16 |
RAOF | Yeah, thanks. | 01:16 |
Sarvatt | you're going to need to force it to evdev for it to work, synaptics wont work unless its a non synaptics touchpad that works with it like a macbook one, seems like a lost cause | 01:17 |
qengho | RAOF: log, conf lacks 50-synaptics. http://pastebin.ubuntu.com/7128307/ | 01:21 |
RAOF | Sarvatt: I think synaptics doesn't report TouchBegin/.../TouchEnd for two-fingers, right? | 01:22 |
qengho | At three, it magically wakes up and starts reporting. It decided it wasn't a scroll request. | 01:22 |
RAOF | Right. | 01:23 |
RAOF | My magic touchpad agrees. | 01:25 |
RAOF | qengho: So, the answer is - ugliness. | 01:26 |
RAOF | I think for the moment you get to assume that you won't get two-finger touch data on touchpads, and give up on getting pinch & such to work :( | 01:26 |
qengho | Is the long answer, kill kill kill the old scroll hack and make every client xi2-aware? | 01:27 |
RAOF | Well, and also the _new_ scroll hack. | 01:28 |
qengho | Oh. I didn't know there was a new one. | 01:28 |
RAOF | Yeah, rather than translating to buttons we translate to scroll axes - which is why you can get smooth scrolling support. | 01:28 |
qengho | So. .....so! I kind of reckon that a UDS immediately after 14.04 is a perfect point to propose this. | 01:30 |
RAOF | But this needs core input support - we can't send both synthesised scroll events and touch events, because clients won't be expecting it. | 01:31 |
RAOF | Can I interest you in getting this to work on Mir instead? :) | 01:31 |
RAOF | Hm. Maybe it wouldn't be _so_ terrible in X. | 01:32 |
qengho | I don't know anything about mir. I'm some browser guy | 01:33 |
RAOF | I think the answer is that it should actually just work in Mir - check out racarr's Chromium-on-Mir work. | 01:33 |
qengho | I keep meaning to merge that and get it into upstream. | 01:34 |
qengho | And, that's good news. | 01:35 |
RAOF | I should _keep_ it working on Mir, too. | 01:39 |
RAOF | :) | 01:39 |
RAOF | Hey! It looks like magic trackpads no longer panic my kernel. Bitchn'! | 01:40 |
qengho | I'm taking credit for that. | 01:43 |
qengho | RAOF: Okay. So, 1a) give up on synaptics pinch. 1b) break the bad news to ara and manager-y people. 2) get mir patches into chromium, first local, then upstream. 3) something about future driver or clients or something that I don't understand and will leave to you. | 01:45 |
qengho | RAOF: thanks for your time. | 01:46 |
RAOF | qengho: No problem! | 01:51 |
qengho | Summary in email to ara. | 01:51 |
RAOF | Which I will be receiving shortly, I guess? :) | 01:52 |
qengho | Now, I must go fall unconscious and hallucinate for 8 hours. | 01:52 |
qengho | Strange we think that's normal. | 01:52 |
qengho | Laters. | 01:52 |
RAOF | Sleep well! | 01:53 |
RAOF | Hey! Looks like Chrome managed to wedge i915 pretty good. | 03:32 |
mlankhorst | RAOF: known bug i think | 08:43 |
=== psivaa_ is now known as psivaa | ||
=== psivaa is now known as psivaa-afk | ||
=== psivaa-afk is now known as psivaa | ||
=== qengho is now known as CardinalFang | ||
=== CardinalFang is now known as qengho | ||
=== qengho is now known as CardinalFang | ||
=== CardinalFang is now known as qengho |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!