subdiff added inline comments. INLINE COMMENTS
> Kanedias wrote in remote_access_interface.cpp:206 > > Can a rogue client do it though? This would crash the server then? > > Yes, I guess so... What would you propose? Should we send it only to first > bound? Or last one? > > P.S. Even more: this interface has no authentication/authorization at all, so > any client can connect and steal our video buffers. > Martin said that first version of protocol can go without it and we can readd > it later (as with fakeinput protocol). Only first bound like you do it now. Just remove the Q_ASSERT (and make sure `boundScreens.size() >= 1`, otherwise continue). > no authentication/authorization at all That's a generic problem yet to be solved on Wayland / the Linux desktop. This also correlates with the push to containerized apps. I would just want something like the permission system in Android, but there might be better solutions. It's a bigger project for sure. Also see here for some early thoughts on it, which to my knowledge until now did not lead to anything more: http://www.mupuf.org/blog/2014/02/19/wayland-compositors-why-and-how-to-handle/ REPOSITORY R127 KWayland REVISION DETAIL https://phabricator.kde.org/D1231 To: Kanedias, graesslin, davidedmundson Cc: subdiff, ngraham, alexeymin, #frameworks, davidedmundson, plasma-devel, leezu, ZrenBot, progwolff, lesliezhai, ali-mohamed, jensreuterberg, abetts, eliasp, sebas, apol, mart, hein