> > Exactly. But I expect other vendors to support JTS so isn't it a good thing to
> > support it in jBoss as well?
>
> Only if we were IIOP/CORBA based, which we're not. To support JTS but
> not be IIOP-based is pretty pointless.
>
Ok.
> > The spec is pretty good along with an understanding of OTS. I do not think it is
> > rocket science either but I really think that it is quite complicated when it comes
> > to supporting transactions that spans more than one TM or even with more than one
> > XAResource (RM).
>
> Why?
Because there are more thing to cover when you more than one TM ordered as a tree of
subordinate coordinators, than when you have a single TM with a few RM's.
>
> > Cause a RM can't help you much for distributed transactions. What I
> > think makes it hard is the OTS concept of "subordinate coordinates" (I think all TM
> > implementations need that, no matter if it is an OTS implementation or another.)
>
> Sure, any decent TM would use subcoordinators. Still not very difficult.
Ok.
>
>
> > and
> > the large amount of possible failures that can happen,
>
> All of which are reported to the TM as an exception.
I don't think that is the case. Consider the case when we have subordinate coordinator
as
a participant in a transaction that dies after sending PREPARED to the coordinating TM.
When the participating does its recovery it see that COMMITED is not logged so it has
no
way of telling if the coordinator has sent COMMIT or not, it then has to run a
termination protocol of some kind. Right?
>
> > still, it isn't rocket
> > science but I still think it is a programming task that required a lots of time to
> > be stable and complete.
>
> Agreed. It would take some time to get it perfect.
Perfect we have agreed on that one ;-)
Still I think much thinking is covered in the OTS spec (and all the XA stuff) but if
much
of it isn't needed in a J2EE implementation, fine.
Regards,
/Tommy
