Deadlock issue in 1.1 (Was: [VOTE] Release Apache Jackrabbit 1.1)

[Jukka Zitting <ju...@gmail.com>]

Hi,

On 10/6/06, Miro Walker <mi...@gmail.com> wrote:
> Yeah - it's a fair point. The problem with the deadlock-related stuff
> is that it's not such an easy job to contribute in this area. The
> locking code is distributed widely throughout the core, and ideally
> what's required is a design rethink for the versioning functionality
> in jackrabbit. While other bug-fixes are quite reasonable to expect
> contributions for, deadlock is almost inevitably going to need to be
> sorted out by the longer-standing developers I think.

Agreed. This strikes close to the issues raised in the recent
discussion on the accessibility of the core codebase. I don't think
there's much we can do along those lines for this issue, but I hope to
have this barrier lowered in future.

> I'm persuaded re: releasing 1.1, though, I'm just concerned that this
> issue is dragging along and I can't see an easy way to get it fixed.
> In our case the only workaround has been to single-thread the entire
> application and put up with total user lockout during long-running
> transactions.

Yeah, I feel your pain. Thanks for bringing this up, I hope we can do
some progress on it and I'll certainly consider doing a 1.1.x patch
release as soon as we have a fix.

BR,

Jukka Zitting

-- 
Yukatan - http://yukatan.fi/ - info@yukatan.fi
Software craftsmanship, JCR consulting, and Java development