Re: Ditching CORBA, again?
- From: Navindra Umanee <navindra cs mcgill ca>
- To: Dan Kegel <dank alumni caltech edu>, Cristian Tibirna <ctibirna total net>, "gnome-kde-list gnome org" <gnome-kde-list gnome org>
- Subject: Re: Ditching CORBA, again?
- Date: Sun, 10 Oct 1999 21:44:22 -0400
Here's one of the articles, published by Sun Microsystems, that
Cristian is referring to:
http://www.cs.mcgill.ca/~navindra/kde-devel/kwn-18/smli_tr-94-29.ps.gz
Summary: Unifying distributed and local computing is simply the wrong
approach and should not be the holy grail of programming.
I'm CORBA-impaired personally, so I'm absolutely talking through my
hat here, but to quote from
http://developer.gnome.org/arch/component/orbit.html:
"CORBA (the Common Object Request Broker Architecture) is a
standard for "distributed objects". This basically means that
applications may make invoke operations on objects that are
not located in the same address space. Frequently object
client and server are in different processes or even on
different computer systems."
"The ORB hides all the low level communications that are
necessary for sending requests to objects, receiving replies
from them, and making object implementations accessable. To
the application, invoking an operation on a distributed object
acts the same as a local function call."
This is exactly the kind of thing that the paper is attacking and what
has obviously struck a chord with the KDE folks. CORBA is still being
retained where it is actually useful and to deal with interoperability
issues; it is simply not the preferred method for the particular task
of embedding applications in local use of the GUI -- Kanossa is
evidently superior for that task.
A middle ground is also mentioned in the paper but I have no idea how
CORBA fits in with this idea. I guess these are the kind of issues
ORBit tries to tackle?
-N
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]