On Thu, 21 Mar 2002, marc fleury wrote:

> |A proxy factory is tied to an invoker. See?
> 
> no, only IIOP,

Well, this is not fully accurate.

> only IIOP because IIOP doesnt generate java clients but IORs... the PF I
> talk about generate java clients with abstracted invokers. In fact one per
> available invoker.

If we assumed an all-Java environment, your very clever serialized proxy 
scheme would also work with IIOP. 

> These can generate JRMP/ClusteredJRMP/WebServices/JMS/ enabled invokers ALL
> transparently.

Can also work for IIOP, in an all-Java scenario. But what would go through 
the wire in that case would not be standard IIOP anymore.

> That IIOP fails to do that (as it failed with the transaction isolation) is
> a IIOP problem,

Not really. Let me clarify this, for the sake of people that might be 
lurking around to absorb technical wisdom from jboss-dev. These folks 
will get the wrong impression that there is something inherently bad 
with the IIOP protocol. There is not.

The problems Marc mentioned arise when you want to support non-Java clients.
Language heterogeneity is the cause of these problems, not IIOP.

> "fuck IIOP"
> -Oberg 2000-

Yes, I've seen that in this list before. Whoever said it could have 
said something else:

   "fuck heterogeneous distributed object computing"

Or, if you prefer:

   "fuck Visual Basic clients calling components deployed in JBoss.net"

It's pretty much the same thing, isn't? 

Best,

Francisco


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

Reply via email to