Hi,

On Mon, Jun 25, 2012 at 8:12 AM, KÖLL Claus <[email protected]> wrote:
> Is it possible to create the Lock on the Global Revison Table only in one 
> phase
> (commit phase) and not spanned over prepare and commit ?

Ideally that should be possible (true two-phase commit), but I don't
think the current Jackrabbit architecture can bend to do that. The
main problem is that the prepare phase needs to be able to ensure that
no concurrent change can invalidate the prepared changes before they
get committed. We currently have no way of doing that without the lock
that spans both prepare and commit.

BR,

Jukka Zitting

Reply via email to