On Tue, Aug 01, 2017 at 12:34:59PM -0300, Guido Günther wrote: > [ cc'ing the gtk+ maintainers ] > > Hi, > On Fri, May 19, 2017 at 01:44:41PM +0200, Guido Günther wrote: > > On Wed, May 17, 2017 at 11:40:22PM +0200, Cyril Brulebois wrote: > > > Package: virt-viewer > > > Version: 1.0-1 > > > Severity: important > > > > > > [ x-d-cc: debian-b...@lists.debian.org ] > > > > > > Hi, > > > > > > Trying to use xdotool (either on the command line or through its xdo > > > python bindings) fails when sending key events to a specific virt-viewer > > > window. In this case, XSendEvent is used instead of XTEST, which doesn't > > > seem to be supported by virt-viewer, and anything sent this way is > > > ignored. > > > > Is there a simple one liner to test this? > > It seems this is not working for any gtk+3 application. While xterm > works as described in the xdotool manpage a simple: > > zenity --entry --title xsendeventtest & > xdotool type --window $(xdotool search "xsendeventtest") asdf > > does not work here either. Maybe the gtk+ maintainers have and idea how > to get this working? > > > > It's possible to raise the virt-viewer window and then send key events > > > without specifying a target window, which is implemented using XTEST in > > > xdotool, but that means messing up with the focus in the surrounding > > > window environment, which is highly impractical. > > > > > > I've also tried switching from default spice to vnc, still no luck. No > > > changes when I tried 5.0-1 from a sid chroot. > > > > > > Any help/suggestion welcome, it prevents me from automating d-i tests > > > with libvirt. (While they do run fine with a “bare” kvm.) > > Can you point me to the code you run the tests with? You should be able > to use the same mechanism with libvirt as without it. > Cheers,
Have you considered using virsh send-key to inject the key presses inside the guest? Christophe
signature.asc
Description: PGP signature
