split from #139: we now handle grabs correctly and simulate pressing the
Escape key when the application loses focus (since there are no grabs on win32).
The problem with the drop down windows as used by chrome is that they are:
transient-forattribute on the temporary window either
I'm not sure there's much we can do about it. Any amount of guessing is bound to break well behaved applications. It would be much easier to get chrome fixed instead.
FWIW: I've tried google chrome 32, and now google chrome beta 33 (which is meant to be using the "aura" toolkit instead of gtk), as well as chromium. No luck with any of them.
Here's a diff of the
xwininfo showing the new drop down window as a child of the root window, it should really be a child of the browser window it came from:
6c6,10 < 110 children: --- > 111 children: > 0x2a50015 "Google Chrome": ("google-chrome-beta" "Google-chrome-beta") 252x302+513+579 +513+579 > 2 children: > 0x2a50018 (has no name): () 252x302+0+0 +513+579 > 0x2a50016 (has no name): () 1x1+-1+-1 +512+578 164a169,172 > 0x2a4a2d3 "Test Page - Google Chrome": ("Google-chrome-beta (/home/antoine/.config/google-chrome-beta)" "Google-chrome-beta") 1515x898+23+45 +23+45 > 2 children: > 0x2a42419 (has no name): () 1507x789+4+105 +27+150 > 0x2a4a2d4 (has no name): () 1x1+-1+-1 +22+44 170,173d177 < 0x2a4a2d3 "Test Page - Google Chrome": ("Google-chrome-beta (/home/antoine/.config/google-chrome-beta)" "Google-chrome-beta") 1515x898+23+45 +23+45 < 2 children: < 0x2a42419 (has no name): () 1507x789+4+105 +27+150 < 0x2a4a2d4 (has no name): () 1x1+-1+-1 +22+44
The drop down window is
252x302 pixels in dimension.
I have created an upstream bug with chrome: chromium #349322
Nothing we can do about this but wait for upstream to fix their code.
Upstream bug has been archived due to inactivity.. might need a prod, else it'll never be looked at.
this ticket has been moved to: https://github.com/Xpra-org/xpra/issues/519