[MUD-Dev] OT: DCOM and RMI

Vadim Tkachenko vadimt at 4cs.com
Thu Jan 8 18:46:24 CET 1998


Jon A. Lambert wrote:
> 
> On  1 Jan 98 at 15:32, Chris Gray wrote:
> > [Jon L:]
> >
> > :I do not rule out ActiveX or DCOM.  However, I have converted my previous
> > :client, which was written using VBScript and ActiveX into Java.  The Sun
> > :vs. Microsoft lawsuit is very much an extension of the DCOM vs. RMI war. As
> > :the agreement between MS and Sun was secret, there is know way of knowing
> > :how this will fall out.  You may see DCOM supported by Java in the end as
> > :well as RMI.  :P  I can well understand why MS refuses to support certain
> > :Java 1.1 extensions.  Oops, I've strayed off signal again...
> >
> > Off topic, but, can someone expand on the acronyms 'DCOM' and 'RMI'? I
> > know Java, and think I understand what ActiveX is (never seen it tho),
> > but I don't know those two.
> >
> 
> DCOM stands for "Distributed Component Object Model".  It's largely
> an extension of the ActiveX components.  Which in turn are an extension
> of OLE, "Object Linking and Embedding".  OLE's earliest and most visible
> use to the average user, was the embedding of Excel spreadsheets in
> Word documents and in embedding other WinThings in other WinApps.
> 
> It exposes an interface of properties and methods residing and executing
> on a server remotely to a client (any C/C++ program) or browser (IE with
> VBScript and/or J++ of course)
> 
> RMI stands for Remote Method Interface

Remote Method Invokation, to be absolutely exact

> which is supported under Sun's
> Java 1.1 which allows one to execute a Java method residing on the
> server.
> 
> DCOM enabled objects are somewhat difficult to write, yet not as
> difficult as CORBA (BTW this is complete hearsay and subjective
> opinion :> ).
> 
> _I_ believe DCOM to be technically superior to RMI in the area
> of performance and features.  OTOH, RMI is quite easy to write,
> to call, and understand which is a very big plus.

[all below is my personal opinion]

I guess, all that DCOM vs. CORBA vs. RMI dispute is just yet another
religious war.
Please refrain from continuing this already off-topic thread as a flame
war, but from my experience (I'm almost the only UNIX freak in a company
which announces itself as a Microsoft Solution Provider) I can state a
few facts:

- CORBA is difficult, but understandable. Also, it's not free (the only
known to me free CORBA implementation is Voyager from
www.objectspace.com (they also have excellent, though a bit
heavy-weighted container library, JGL).
- RMI, as strange as it sounds, is a lot like CORBA, but it feels much
better in Java environment. The disadvantage (once again, for me) is
that I've been developing everything (including commercial applications)
under Linux, which meant much less support for everything commercial
and/or proprietary (JDK, Netscape with JDK 1.1 support (btw, released at
December 24)) - you depend on raw implementations - Java is not as
mature as I want it to be, but mature enough to see it as a great
opportunity to write a reusable code, thus speeding up the development.
- COM/DCOM/ActiveX is a technology strictly bound to Windows (and don't
pay attention to Microsoft announces to make it cross-platform - I don't
want to elaborate on this, the facts speak for themselves), which is not
an option to me - see, I've been to MS-DOS, then Windows 3.1, then OS/2,
then UNIX with Windows 95, now I'm on UNIX and pretty happy with that.

The bottomline - try to be protocol-independent. I've just implemented
the reusable client/server model which allows you to implement just a
business logic and select whatever protocol[s] is/are suitable for a
specific application and will try to stay with it.

> Jon A. Lambert

--
Still alive and smile stays on,
Vadim Tkachenko <VadimT at 4CS.Com>
--
UNIX _is_ user friendly, he's just very picky about who his friends are



More information about the mud-dev-archive mailing list