Hello,

-----Urspr�ngliche Nachricht-----
>Von: Ole Husgaard [mailto:[EMAIL PROTECTED]]
>Gesendet: Donnerstag, 2. August 2001 11:49
>An: [EMAIL PROTECTED]
>Betreff: Re: AW: [JBoss-dev] jbosspool: implementation of
>connectionClosed() in org.jboss.pool
>.connector.BaseConnectionManager$XAListener


>Hi,

>but will result in TxCapsule
>doing start(TMJOIN) instead of start(TMRESUME).

that�s what I tried to contradict with the code segment.
 
And that�s where I have to retract my opinion and ask myself, why the hell 
I did not see that TMJOIN thingy in the debugger ... I have been convinced
not to get the start() call
at all. Must have been a hallucination  or somewhat. I�m going to check this
... 

>Problem here is that I am not sure if the JTA spec
>wants this to be allowed: If it is allowed, there
>is no need for the TMSUSPEND and TMRESUME flags, as
>you could just use end(TMSUCCESS) instead of
>end(TMSUSPEND), and start(TMJOIN) instead of
>start(TMRESUME).

>Tecnically JTA does not forbid it, but why did they
>introduce TMSUSPEND, TMRESUME and the suspended
>resource state then?

>To be tolerant, I made TxCapsule allow it.
>But in case some resource does not allow it,
>it is better to use delist(TMSUSPEND) and enlist()
>than delist(TMSUCCESS) and having enlist() fail due
>to resource failure.

I�m not sure what would be the "right" approach, still feeling that
TMSUSPEND would be the 
close-to-specification choice. It�s a single liner in either my resource�s
start() or the
XAListeners.connectionClose() method.

Thx much for your opinion, guys.
CGJ

_______________________________________________
Jboss-development mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/jboss-development

Reply via email to