> Has all this been committed to head ??

The new XA functionality is there, however it was not tested, especially
your case. So it might work, might crash. At the moment one XA-related
test fails, so no guarantees.

> Would be very willing to try out the new XA code and do some testing.
> The current code does some very strange things across our cluster at times.

Same symptoms? More details?

> After our last discussions about the XA code and not to use head, I thought I would
> wait a bit until the code stabilised a bit.

Well, it is not yet stabilized... it is quite hot development right now.

> By the way is in the old code is access to the tables holding
> The XA transaction data in some way serialised ?

Nope, unless you specify the custom TPB.

> When we get a failure with
> timeouts in the XA transactions it seems to affect the whole cluster simultaneously
> rather then just one machine and the machines are not sharing the smae transaction.

What does "affects" means? Can it be just "normal" Firebird's behavior?

As of today:

>> - new XA functionality (done, requires testing);

in HEAD

>> - Java 6.0 support (I have the code already, will commit it soon,
>> testing needed);

basic code in HEAD, few methods must be implemented before release

>> - improved Hibernate support (nothing's there);

committed yesterday - hopefully now the correct exceptions will be
reported. BTW, can somebody prepare a test case that would check whether
Jaybird does it correctly?

>> - two small fixes that were contributed by Evgeny Putilin privately;

Not yet.

Roman