I've been following of the TC4.0/TC3.2/TC3.3 threads for a while in
silence, and wanted to make an observation.
The Tomcat project really has two major types of customers: those who
want an open source servlet container for their production needs
("conservatives"), and those interested in the state-of-the-art
("tinkerers"). The project has made the decision--by supporting both 3.x
and 4.x in parallel--to try to serve both (a good decision, since I think
there are plenty of people in each user group).
The values of the conservatives are
* stability
* lack of bugs
* a seamless install
* performance
The values of the tinkerers are
* compliance with the latest standards ("100% buzzword compliant")
* advanced features
* elegance
These two groups have very little in common, except this: both will
tolerate change so long as it's in the direction of their values. A
conservative will not argue with you about changing Tomcat 3.2 to 3.3 if
it's to make it faster, more stable, or less buggy. "More elegant" or
"more featureful" doesn't cut it, though, which is why I suspect there
will be little interest in Servlet 2.3 on top of 3.x (thus it's good for
it to be a "contrib" element, for the few who DO want such a combo).
I think so long as 3.3 is a move--as Costin says it is--toward a
better-factored, less-buggy version of 3.2, and all the changes taking
place are just moving around established, tested code, then it's a good
direction. I have consulting customers who may very well end their project
lifecycles still on servlet 2.1, old school or not!
So long as we keep in mind who's served by a change, and keep it
consistent with what that groups wants, I think we can safely move 3.x
through a fairly long arc. There seems to be enough interest in each
viewpoint to support 3.x and 4.x at once without bogging down the project.
regards,
kd
