https://bugs.kde.org/show_bug.cgi?id=522899
Kevin Krammer <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |[email protected] --- Comment #1 from Kevin Krammer <[email protected]> --- This is not an expert analysis, I was just curious on how this was implemented. I think the main issue is that XwaylandLauncher handles problems of the Xwayland process, e.g. crashes, in XwaylandLauncher::handleXwaylandFinished() but QProcess::finished() is not emitted when the process couldn't be started at all. So the cleanup code in XwaylandLauncher::stop() is never called. I guess the "can not even launch" case needs to be handled in XwaylandLauncher::handleXwaylandError() instead. Additionally, XwaylandError::enable() could error out early if the Xwayland executable is not there at all, e.g. return false. Xwayland::init() could then also error out early, e.g. also return false, likewise in ApplicationWayland::performStartup() or exiting at this point -- You are receiving this mail because: You are watching all bug changes.
