I suspect it might have something to do with timing. The window's "destroy" signal invokes aplwrap_quit(), which then terminates the APL process.
The termination of the APL process invokes (due to the g_child_watch_add() call) apl_exit(), which then tries to terminate the APL process. I suspect that, depending upon the details of one's kernel, scheduler and GTK+, it's conceivable that (for the sake of supposition; I have no concrete evidence this case could actually work) the window-destroy path could complete and let aplwrap terminate before the secondary apl_exit() call ever gets a chance to run. That much seems plausible. (To me, at least.) What I can't grok is how the clipboard content has anything to do with the termination behavior. Maybe there's an extra content switch to send the clipboard to the GNOME shell...? On Fri, 2014-09-12 at 15:30 -0400, Chris Moller wrote: > I just applied and pushed David's patch, the question remaining why it > showed up only in some environments and not others.
