On 06/03/2015 12:54 PM, Stephan Bergmann wrote:
On 06/02/2015 02:32 PM, Lionel Elie Mamane wrote:
So, if you would like to work on that (which would make me happier),
then apply the attached patch, then:
make JunitTest_dbaccess_complex
several times in a row, and you should get the problem. If not, try
uncommenting more "@Test" instances.
Thanks for the reproducer. Looking into it a bit, it smells like a
reincarnation of an ages-old phenomenon where the JVM garbage-collects
objects "too aggressively" (i.e., with the object still nominally on the
call stack, but optimized away) and we thus send a release call from
Java URP to binary URP too soon. I'll look into it further.
Ah, not quite. Should be fixed now with
<http://cgit.freedesktop.org/libreoffice/core/commit/?id=3766f2a0d3bc50919366b0e43f5ee8c814019c8b>
"Finalizers must not access UNO objects" (see there for details).
_______________________________________________
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice
