Created attachment 55883 [details] screenshot When running winefile (wine-1.3.36), I can see gradient artifacts in the window title bar of internal (emulated) windows that are not present when running the same test with UXA. Steps to reproduce: - Start winefile - De-Maximize the current file-view - the current view should now be an emulated window, its gradient should show the artifact My System: i945GM intel-2.17-git
And I thought you were going to have noticed the swaths of purple...
I don't have the same gradient for my window title bar. Hmm, presumably I could replace my *.reg with yours to get an equivalent setup?
Created attachment 55898 [details] screenshot with two equal gradient colors
Created attachment 55899 [details] wine registry files
hope I done't bother you with this kind of reports - actually the wine reports have no real value for me... attached are my registry files.
The wine reports are interesting as they appear to be a stress test of code paths rarely used elsewhere. And if you haven't noticed, running 2D benchmarks under wine is faster than running than under native win32. Of course I'm sure a large part of that is that wine is merely approximating the look of the windows functions. But it made me feel better ;-)
Hmm, set the gradients by hand and tried your *.reg (thanks!) on my PineView box. Seems ok. Either it was a further side effect of (sna: Give kgem_create_buffer_2d a unique id), which is not impossible, or 945gm is broken. The latter is much less likely since pnv is just 945gm on die with a few minor tweaks, but none as far as I am aware in the render engine.
I was testing with the unique-id patch already in - so it seems to be something different - strange. Maybe my system is somehow broken?
At this moment in time, it's more likely to be a difference between our systems that I haven't factored into the code. I'd like to bring up my old 945gm to cross-check the rendering before declaring I have no idea what's going on ;-)
Note to self, we discovered that the change to the new DIB engine triggered a few bugs elsewhere so that may be relevant. Clemens, you don't know whether wine-1.3.36 included the new engine? I'm waiting on debian packaging wine-1.4 before retesting on PineView. I should just go and dig out the 945gm as well just to be certain.
Ok, on the 965gm with an updated wine and having converted the gen4 backend to also use the gradient ramp, I can reproduce the issue. The implication is then the gradient approximation is not robust enough for this use-case.
commit 5c4dc9c5db7b2a5b936bdbc15536c5cf0f7f5f23 Author: Chris Wilson <chris@chris-wilson.co.uk> Date: Mon Apr 2 13:40:22 2012 +0100 sna/gen3+: Fix sampling of borders around gradients Incurs a slight loss of precision for the internal gradient, but much more preferable to the artefacts around the borders with RepeatNone. Reported-by: Clemens Eisserer <linuxhippy@gmail.com> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=45016 Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.