Marc Haber wrote:
Is there any way to reproduce this without deliberately crashing an X session?
Probably. Reading the bug report, it sounds like there is a known race condition that may cause this. I suspect that whatever sequence my X clients died in hit the race condition. The xterm probably started to exit, leading to reads failing. debfoster probably took the read failure as me hitting enter (e.g., the default) and kept going. Then it probably got a SIGTERM.
I'd guess that an appropriately-timed SIGKILL could reproduce it. Or maybe just hitting the close box in the xterm its running in.
Asides from that, I agree that debfoster should write new keepers to a tempfile.
I think that the ftruncate in writekeepers is very likely the cause of this bug. Fixing that function to do things the Right Way(tm) with a tempfile and rename(2) should fix it.
Greetings Marc
-- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
