Comment 26 for bug 786317

Revision history for this message
Christopher Small (metasoarous) wrote :

I did some more research and realized a few things.

Something I've failed to mention till now - I'm running 11.10 AND Gimp GIMP 2.7.3 (a current snapshot of development towards 2.8). Although there were definitely still issues (at least as I recall) with this version of gimp before the Ubuntu upgrade, after playing around with nearly all of the UIs on my computer, this combo seems to make things a lot better, at least usable, even in 3D Unity.

I ran through Gnome Classic (with effects, though there weren't really any actually enabled, so the test coverage could be improved upon here), Unity 2D, Unity and Gnome Shell. Of these, I was only able to produce issues with Unity 3D. And as I've discovered, even that seems to be much better now; at least usable.

I opened a blank canvas and did rapid bunches of four strokes at a time so that it would be easy to count misses. I had zero problems with any configuration except for Unity 3D. I did 36 bunches, and only had a total of 18 missed strokes. What is interesting is that once I hid the toolbox (by pressing Tab - not sure if this functionality was present in Gimp 2.6) the misses went down to 0 out of 60 bunches. Running my wacomsettings script (with xsetwacom calls - see an earlier post of mine) didn't seem to have any negative effect here. Once I brought the toolbox back to the front, I experienced only one or two misses (out of sixty bunches, if you can believe it). Not sure why the behavior didn't come back as strongly, but it didn't. This is already a significant improvement from the sticking-every-other-time behavior I was seeing a while back. And I actually really like being able to hide the Toolbox while I'm drawing (except when I have another screen I can let it sit on), so that to me is a good workaround. And as I said, even with the toolbox up, the bad behavior seems far less frequent now. So, I definitely recommend everyone try out Gimp 2.7 to see what change in behavior you get from it.

I think that my results definitely further support the hypothesis that the Toolbox and window stacking is somehow part of the problem here.

A few more notes about my setup for the sake of thoroughness. I'm using an Intuos 4 and the new Wacom Graphics Tablet GUI within System Settings to set the Tracking mode to relative (which I kind of like, since it doesn't skew the tablet strokes when I'm using multiple screens). I played around with that a bit while testing out Classic Gnome, but it didn't change anything. I should have tried it in Unity to see if that made things better or worse (since that is the one component that was actually having any issues for me). That is an area that could stand more testing. If I get the time to, I'll definitely share my findings. It might also be worth testing out the effects of changing parameters on the mapping matrix a bit. I had angle mapped to tilt, color mapped to pressure, and hardness to force. I had size mapped to velocity at one point, and it seemed that I was getting a slightly higher rate of issues there, but that was before embarking on my "bunches" testing (without which, it's really hard to get a good sense for how bad things actually are).

I'll also point out that whereas (as I mention in my earlier post) I did experience some problems just after running my wacomsettings script in Gnome Shell, I didn't this time around. That might have been a fluke. I'll definitely keep an eye on that. At this point I intend to mostly use Gnome Shell, even though it's a bit buggy still.

On a final note, I didn't see any of the problems which Andrew has described, but obviously there are a lot of parameters to tweak between our two setups where one might find issues.

Cheers

Chris