[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-core-devel
Subject:    Back to the basics
From:       Kurt Granroth <kurt_granroth () email ! mot ! com>
Date:       1999-09-29 16:06:53
[Download RAW message or body]

I've been reading this debate on Canossa vs CORBA and was a bit
suprised on the tack that it has taken.  Waldo, David, et al -- I
totally understand your position and your concerns.. but I think that
you are straying a bit from the fundamentals that KDE is built on!

Much has been written and said on the differences between KDE and
GNOME, but in my mind, there exists only one CORE difference.  GNOME
will always use the One True Solution to everything -- even if it
means sacrificing everything else.  KDE has *always* been completely
practical.  We are driven by three core principles:

1) REUSE as much code as possible
2) The code must be STABLE
3) The code must work RIGHT NOW

The project founders went with Qt right from the beginning knowing
that they were opening a licensing can of worms.   Why?  Because Qt
fit well with the core principles.

Back to CORBA.  CORBA is clearly the One True Solution.  It is
tremendously powerful as a buzzword and I think we should keep using
it *somewhere* in KDE just so we can include the phrase in our press
releases.  However, it is currently violating almost everything KDE
stands for!

We are NOT reusing the maximum amount of code as possible (although
using mico instead of writing our own ala orbit is better then
nothing).  David, you mentioned that you are willing to reimplement
the entire Qt toolkit in OParts.  WHY??  Qt is already implemented!
Wouldn't you rather spend your time implementing cool new features
instead of simply playing catch-up all the time?

The code is NOT stable.  You all know about our embedding stability
problems so I won't go into that much more.  You can argue that if we
continue working on it, it will eventually become more stable.. and
that would be right.  But WHY do that if canossa is *already* stable?

Our CORBA based OParts has the one redeeming factor that it does work
right now... albeit not tremendously well (see above).

The two arguments in favor of using CORBA are so:

1) An open interface allows projects like GNOME to use our components
   (and vice versa)
2) CORBA allows for networked components

I reject 2) completely.  I have not ever heard of this happening.

The first one is more tricky.  Some of you probably know that I am a
big supporter of integrating GNOME into KDE as much as possible.  As
such, I agree that the IDEA of sharing components is very attractive.

However, this line of thought also scares me a bit.  Have any of you
seen a single instance where the GNOME team has tried to work with
OParts in the year that it's been around?  NO!  In other words, this
is all just a mind-exercise.  It's a great idea that might or might
not happen and it's fun to sit around and TALK about it... but it's
not a reality right now.

Can I remind you all of our core principles?  Reusable, stable code
that WORKS RIGHT NOW.  Any collaboration CORBA code between GNOME
and KDE is currently 100% vaporware -- it doesn't exist and we have no
guarantee that it will exist.

So why are we considering hamstringing ourselves on the POSSIBILITY
that GNOME might want to use our components?

To summarize: It looks like canossa fits our core principles quite
well.  I am 100% for it!
-- 
Kurt Granroth            | granroth@kde.org
KDE Developer/Evangelist | http://www.pobox.com/~kurt_granroth
         KDE -- Putting a Friendly Face on Linux

[prev in list] [next in list] [prev in thread] [next in thread]