On Sat, 11 Jan 2020, Marvin wrote: [...]
Full results can be found at: https://testbot.winehq.org/JobDetails.pl?Key=63133
Your paranoid android.
=== debian10 (64 bit WoW report) ===
dinput: mouse.c:152: Test failed: Failed: 00000001 mouse.c:158: Test failed: Failed: 80070005 mouse.c:161: Test failed: Failed: 00000001 mouse.c:164: Test failed: Failed: 80070005 mouse.c:178: Test failed: GetDeviceData() failed: 00000000 cnt:0
Based on the test.winehq.org history these tests started failing on the 26th of December:
https://test.winehq.org/data/tests/dinput:mouse.html
A bisect shows they are caused by this commit:
commit cfcc280905b7804efde8f42bcd6bddbe5ebd8cad Author: Rémi Bernon rbernon@codeweavers.com AuthorDate: Thu Dec 26 12:51:23 2019 +0100 Commit: Alexandre Julliard julliard@winehq.org CommitDate: Thu Dec 26 21:56:04 2019 +0100
winex11.drv: Send HTMENU instead of HTCAPTION to query window activation.
Commit 71d35d8940118bc6de6522913fb8c473fa5b2c24 broke the way WM_TAKE_FOCUS protocol is implemented: WM_MOUSEACTIVATE now replies MA_NOACTIVATE by default when using HTCAPTION.
We use the WM_MOUSEACTIVATE -although Windows does not- regardless of the way focus is changed to check whether a window wants focus, and Windows sometimes changes focus regardless of the message reply.
Steam and the Wine system tray are affected for instance.
Signed-off-by: Rémi Bernon rbernon@codeweavers.com Signed-off-by: Alexandre Julliard julliard@winehq.org
There are actually two versions of this patch on the mailing list:
* https://www.winehq.org/pipermail/wine-devel/2019-December/157117.html - Modifies dinput:mouse, marking the failing tests as todo_wine. So the TestBot found that dinput:mouse succeeded, although with todos: https://testbot.winehq.org/JobDetails.pl?Key=62569
* https://www.winehq.org/pipermail/wine-devel/2019-December/157099.html - Only modifies winex11.drv. Because of that the dinput:mouse test was not run and the regression was not detected by the TestBot: https://testbot.winehq.org/JobDetails.pl?Key=62502 - Posted a couple of days before the version above, this is the version that got committed.
So the patch was known to to cause these failures and presumably that's being worked on (if not it would be good to make a bug report for reference).