Re: More HTTP/IIOP/RMI/etc..

I Find Karma (adam@cs.caltech.edu)
Wed, 9 Apr 97 01:38:20 PDT


> RMI, in particular, is causing confusion in the market, Linthicum
> said. "You see a lot of 'RMI vs. CORBA' messages in the newsgroups,
> where you've got your Java bigots and your CORBA bigots and people
> like me who are both Java bigots and CORBA bigots trying to figure
> out what to make of it all."

Isn't this like comparing apples and orangutans?

> "I believe 1997 is the critical year for CORBA to cross the chasm,"

As was 1996. And 1995. And 1994. Heck, I remember people saying this
about CORBA at ObjectWorld in 1993, before 2.0 came out as a spec.

> Rymer's point was that this may be the make-or-break year in which
> CORBA vendors will either make the necessary investments in
> easy-to-use tools that will bring their technology into the
> mainstream or lose their chance at acceptance. They have to act in
> the first half of 1997, while Microsoft's distributed object
> computing technology is still too immature and incomplete for
> enterprise-scale systems, he said.

First half of 1997 is what, 2 more months? No way this is gonna happen.
And Microsoft isn't naive; they know how to position DCOM.

> Meanwhile, CORBA is emerging as a platform that integrates with many
> other types of middleware to provide a comprehensive solution, she
> said. "But there is a terrible dearth of products that can help you
> build these applications."

You got that right.

> Tim Berners-Lee, the creator of the Web and director of the World
> Wide Web Consortium, gave a keynote speech extolling the virtues of
> combining distributed object technologies with the Web. He said the
> two technologies will continue to follow distinct paths, however.

I'm not convinced. I'd give them about 50-50 odds of continuing along
distinct paths.

> At last year's Object World West, Netscape technology vice president
> Marc Andreessen predicted that IIOP will replace the Web's Hypertext
> Transport Protocol.

And of course, MarcA is never wrong.

> "They've both just moved to version 1.1; they both have problems,"

Everything in release 1.x has problems. To get rid of the problems,
they have to rename the thing to be 3.x (unless the product in question
is Word, for which only version 5.1 for the Mac is acceptable)...

> "What will happen if we get it right is that there will be sharing
> of things on many levels," Berners-Lee said. On the other hand, he
> outlined what will happen "if we get it wrong." In that case, he
> said, a Web page that is displayed according to one system of
> security and authentication might launch a Java applet that would
> use entirely separate and inconsistent security mechanisms to
> communicate via RMI.

Think of it as a separation of church and state. Whether we like it or
not, these efforts may be on a collision course, but it seems to me that
TimBL is interested in meta-questions (protocol consistency, security,
extension architectures, metadata structures, etc) whereas the OMG is
interested in building a middleware bus that is usable in enterprisewide
and cross-enterprise object solutions. If those two goals are
reconcilable, a merger is likely. If they aren't, then CORBA/IIOP will
service one group, and Web Protocols will remain an experimental medium.

----
adam@cs.caltech.edu

The moral is obvious. You can't trust code that you did not totally
create yourself.
-- http://www.acm.org/classics/sep95/