[Xquartz-dev] Fully self contained Xquartz.app?

Eeri Kask Eeri.Kask at inf.tu-dresden.de
Thu Jul 15 02:44:17 PDT 2010


Am 07/15/2010 12:42 AM, Jeremy Huddleston <jeremyhu at apple.com> schrieb:
> On Jul 14, 2010, at 02:45, Eeri Kask wrote:
>> (Because of the latter (window entry/exit and focus events most
>> troublesome) it is hard to use Xquartz for own X11-applications
>> development because nobody knows where to look for problems.  In
>> this context xcb can be questioned too of course.)  :-(
> 
> I don't think I have an open bug report about these entry/exit/focus event issues...


No, definitely not by me.

I haven't yet managed to gather sufficient information about some
regular pattern how these enter/leave/focus issues can be triggered
deterministically willfully.  Currently it looks like it is kind of
understanding some stochastic process by observation...  :-)

The most I can say today, it seems window stacking order changes or
unmap/map requests result in effects which can be explained like the
misordered delivery or dropping of some enter/leave events, which
then understandably often results in a mess in focus management; let
it be focus by PointerRoot by the X-server or by assigning by some
window manager implementing sloppy-focus-follows mouse.

If enforcing stacking order change by some special client
application tied to some "passive-grab/passive-release-grab" hotkey
then it looks like the execution of lowering the affected window
sometimes (i.e. again randomly; not always) finishes first and then
PointerRoot focus (or even enter-window) assignment by the X-server
still uses the stacking order of the pre-lowering screen state or
something... resulting in focus falling into a window which is not
ontop.  (In few cases the server even has reported the pointer being
in a window which is perceptually verifiably covered by some other
window.)

Therefore all in all now I am only able to complain but still keep
observing and try creating a reproducible deterministic triggering
of some of these situations... prior to reporting a bug.


E.g. exposure events cause trouble too, esp. in connection with the
XShape- and AppleWM extensions; but this results only in wrongly
painted screen content, not ending up typing into a wrong window.

(P.S.  If it is relevant, I run X in fullscreen mode all day (not
using click-to-point focusing), returning into Aqua only on logout;
so how far might possible Aqua involvement go in these problems is
uncertain...)

Greetings,

    Eeri Kask



More information about the Xquartz-dev mailing list