Hey
Scott Seligman wrote:
> A single value for this property, and a single implementation of the
> corresponding class, should suffice for the entire Container. Each
> EJB can have a value buried by the Container in its thread-local
^^^^^^^^^^^^
We kinda knew this already. It's what we've (or at least me) been
whining about from the beginnning, that we do *not* want to use
thread-locals. When suggesting a solution to this problem do not include
thread-locals. We do not want a solution, but The solution. Using
thread-locals in this case is closer to *a* solution.
> Note that a few months ago there was some discussion about having
> multiple "loosely coupled" Containers running with a single JVM. In that
> case it becomes a bit more tricky for each Container to supply its
> own URL context factory implementation, and we discussed classloader
> tricks that -- while a bit ugly -- would accomplish this. But that
> discussion was not about the common case, and should be of interest
> only to a few Container implementers doing uncommon things.
Admittedly, but still a valid point. What I have said is that it should
be *possible* for people (like me, who kinda likes loose coupling and
such) to implement J2EE in that way. There are no other big roadblocks
to hinder this other than this.
Thanks for your reply though! But we need more! :-)
regards,
Rickard
--
Rickard �berg
@home: +46 13 177937
Email: [EMAIL PROTECTED]
Homepage: http://www-und.ida.liu.se/~ricob684
===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff EJB-INTEREST". For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".
