Hi,
On 12/13/06, Przemo Pakulski <[EMAIL PROTECTED]> wrote:
I'm mostly concerned about JCR-546.
It seems that proposed patch does not resolve the problem, because it
breaks order of acquired locks (see JCR-632),
You mean JCR-672? As I just commented on the issue, I don't think the
problems are related. JCR-546 can mostly have enlarged the window of
opportunity for this deadlock to occur by extending the scope of the
write lock.
It would be very nice to have a fix also for JCR-672 included in
Jackrabbit 1.2, but I'm not making it a blocker since it's still not
the only open issue on concurrent versioning operations (JCR-18 being
the core known issue).
In long term it seems that all usage of locks needs to be identified and
reviewed again.
Agreed.
BR,
Jukka Zitting
