axis-java-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sam Ruby" <ru...@us.ibm.com>
Subject Re: WASP & Axis
Date Tue, 04 Dec 2001 01:33:54 GMT
Radovan Janecek wrote:
>
> Hello all,

Hello Radovan!  Long time no see!

> However, I feel there are other serious questions we should discuss prior
> doing any merging effort:
>
> timeframe

It is hard to discuss timeframe without seeing the code.  In a perfect
world, one would be obviously better and plug compatible; but clearly that
won't be the case so the discussion is likely to be messy for a while.

> ways to cooperate
>
> Since I think that Axis and WASP are pretty incompatible, I definitely
> prefer the third option. I'm convinced that it is the best one at least from
> the technical point-of-view. Is the 'Xerces and Crimson' case a good example
> of the third option?

Xerces and Crimson is a perfect example - both of what we want to avoid and
what will ultimately be the right course.

For the longest time, Xerces and Crimson had an uneasy coexistance
externally, but internally, they had a bitter relationship.  Essentially,
all of the developers of Xerces were from one company, all of the
developers of Crimson were from another.  Both sides could provide ample
"evidence" that it was the other side's fault.

Fast forward to the present.  Xerces 1 and Crimson are both essentially in
maintenance mode at the present time.  The active code base is Xerces 2.
The new design was jointly produced by a common team.  The implementation
bears little resemblance to either code base - but incorporates important
lessons learned from both.  The one thing that was vitally important was
that the name of the code base with the most name recognition associated
with Apache lived on.

 = = =

We broke that rule of successors keeping the same name here in Axis, and we
have suffered for it.  I don't know if you recall, but I argued weakly for
keeping the name SOAP, but agreed to abide by the consensus.  For the
longest while I argued strongly that we should call this release V3, but
over time, stopped that.

I take responsibility for allowing this rule to be broken.  The name SOAP
had little support from the PMC, by the development community, or even by
the W3C at the time.  But in hindsight, I should have stood firm.

When you stop and think about it, it is silly that we as professional human
beings assign so much importance to the name, but again and again
observation of behavior indicates that the name is the most important
aspect of community.  Just like people will go to war over flags, you will
see that there are people here who cling on to the notion that there is a
separate SOAP community.

 = = =

It pains me to bring up the above as it is likely to reopen old wounds.
But the lesson is clear.  The successor to Axis should be named Axis 2 (or
if I had my druthers, Axis 4 or perhaps even SOAP 4).  From a technical
perspective, the code base may bear little internal resemblance to the
current code base named Axis.

Another example that we can draw upon - the "Catalina" revolution in the
Tomcat base.  It originally resided in the same CVS tree, got split out to
a separate cvs tree, and anointed with the name Tomcat 4.  Now there is
some activity factoring out common code into a "commons" area.  If we
follow down similar footsteps here, I see the possibility of there being a
separate "Web Services" project, with multiple subprojects focusing on
individual areas.

> WASP for C++
> ----------------
> You probably know that Systinet also has a WASP for C++ too. It has the same
> architectural concepts as WASP for Java. I think it is perfect candidate for
> opensourcing on Apache as well. What do you think?

There is high demand for a cross-platform C/C++ implementation of SOAP.  It
would likely do very well here.  And if it's architectural concepts were
similar to the Java implementation that you are proposing

- Sam Ruby


Mime
View raw message