Hi,
first: thanks for all the valuable information.
My (personal) impression is that IBM prefers RMI over CORBA and I would
be happy if I could choose RMI to solve all my problems.
But there are at least two reasons to choose CORBA:
1) adding objects/services written in other languages to the
(distributed-) system.
2) sending request via Internet - I really can't understand why there is
still no RMI-firewall-proxy for RMI (JRMP) available. If Sun (and IBM)
want to make RMI-based systems usefull in the internet then they should
offer this.
I like the Sun statement "the network is the computer" - but at the
moment the network ends at the proxy-server of your company...
what about performance and stabilty issues ? My impression is that (for
example) the visigenic/inprise visibroker implementation "looks" a
little more scalable and professional - any experiences (measurements)
so far ?
thanks for any comments on this,
sven
Rainer Kerth wrote:
>
> I would not be too optimistic about the possibilities of switching the
> transport infrastructure transparently. RMI and CORBA differ in several
> important ways:
>
> - IDL, if compared to RMI, gives you only a rather limited interface to your
> object (e.g you can't write a distributed propertyChange method with a *real*
> Java event parameter in CORBA - you have to use a struct.). In RMI you can use
> all Serializable types in Interfaces.
> - CORBA does not allow for downcasting of remote stubs. (you have to use
> narrow, and you loose the object identity in doing so). RMI does.
> - CORBA does not guarantee the object identity of remote references (i.e. two
> references to the same remote object need not be identical wrt ==; you have to
> use isEquivalent). RMI does.
> - CORBA does not allow for overloading of methods. RMI does.
> - CORBA cannot transmit null parameters. RMI can.
>
> CORBA is certainly a great way of doing distributed computing, with a long
> (well - relatively long) history of successfull implementations. But you have
> to pay a price for language independence. RMI over IIOP is only in beta so far.
> It is interesting but the points mentionned above concerning downcasts and
> object identity apply to RMI over IIOP as well.
>
> IMHO it will be rather difficult to port an existing RMI application to CORBA.
> If you want to go for CORBA, do it right away. Wrapping the CORBA code in a
> thin Java layer is possible (we did it in our current project) and gives you
> some nice features. It would probably require a considerable amount of work,
> though, and it is not a standard solution.
--
-----------------------------------------------------------------------
Sven-Uwe Meyer
Gruner+Jahr Electronic Media Service GmbH tel. : +49-40-370 377 29
Am Baumwall 11 fax. : +49-40-370 377 57
20459 Hamburg
Germany email : [EMAIL PROTECTED]
-----------------------------------------------------------------------
===========================================================================
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".
