https://bugs.freedesktop.org/show_bug.cgi?id=66025
--- Comment #40 from David Bolen <db3l....@gmail.com> --- Created attachment 82928 --> https://bugs.freedesktop.org/attachment.cgi?id=82928&action=edit Patch without use of sentinel (In reply to comment #37) > Another thing that made me wonder is: I assume that _uno_import will never > be called by multiple threads in parallel, anyway, or else the "d[x] = > unknown" and "del d[x]" might introduce a new race there. It should be safe - the Python import lock prevents two threads from importing simultaneously (and possibly seeing partial modules). In retrospect, it appears more complicated than necessary, and your uncertainty could be shared by future viewers. A simple flag set at the bottom of each exception path works just as well. This patch (via git format-patch) uses that approach. It's off of the current master and incorporates the patch2 change since that wasn't there yet). I'll look into gerrit for the future when I have a chance. -- You are receiving this mail because: You are the assignee for the bug.
_______________________________________________ Libreoffice-bugs mailing list Libreoffice-bugs@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice-bugs