> Ok, I just got off the phone with Pier (who just broke both of his thumbs
> skiing this weekend!!!! ouch!!!!) and he has clarified a bit of what Costin
> expressed to him about what he really wants to do, so let me repeat it to
> see if I got it right now:
Thanks, Pier :-)
> Costin wants to continue to work on 3.x HEAD and do another release that ads
> security and performance enhancements. Then, according to Pier, after that
> has been released, he was going to fork the code and move off to source
> forge and continue on with his development.
I think there is a bit of misunderstanding here - I said clearly I don't
want to do any fork or move off to source forge.
After 3.3 is released I'll continue to do bug fixes, but I plan no major
development on tomcat core. From this point I'll only work on modules and
code that can be built _on_top_of_ tomcat 3.3.
What I said about sourceforge is that I'll develop some of the modules
outside tomcat - and I'm talking about the module that supports
Servlet2.3, and any other "controversial" module.
Again ( and again ): the tomcat3.x architecture is based on a small core,
and modules. All the functionality is implemented in modules - you don't
like a feature you just remove it from server.xml.
Think of Apache1.3 and mod_perl, mod_php, mod_jserv, mod_jk, all the
authentication modules, all the logging modules. They all rely on a stable
API ( coresponding to tomcat core). You don't like a feature - you don't
use it, and the apache release is not suffering from featurism for this
very reason.
In fact the reason Apache1.3 is so successfull is that it can be extended
in so many ways - I think the key for the future of tomcat3.3 will be the
stability of the core, that will allow people to build on top of it the
same way they do build on top of apache.
That's a clear argument _not_ to do any fork or any future change in the
core - except bug fixes and maybe well-justified and backward compatible
changes.
Tomcat3.3 will finally make this clear separation, and will
mark the end of the code refactoring. Tomcat3.1, 3.2 are incremental
steps, where we tested the model ( by implementing features like security,
increasing performance ), improved stability and resolved some of the
bugs.
I offered weeks ago to throw away tomcat3.3 and re-do all the changes
needed to finish this separation out of 3.2.x. Given the number of changes
( almost the same kind as 3.1-> 3.2 ) it would be cheating ( and
confusing ).
> I am +1 on the above.
> I'm -1 on a 3.3. I would call this version something under 3.2.x.
> I'm -1 on adding new functionality to 3.2.x in favor of adding and improving
> on it in 4.0.
As I said, all functionality will be part of independent modules - and I
do plan to re-use any code that is good in 4.0, and add more features on
my own.
But given your -1, none of that will be posted on Apache ( until you
change your vote).
It seems many people agree with you and don't want more features in tomcat
- and I understand that. But you are free to ignore the modules I'll be
writing, and if "duplication" is your concern - try to understand the
word "reuse".
Some of the features I plan can be _reused_ in 4.0 - some can't ( because
of the architecture ). It's not my fault, and if they want (can) reuse the
features - I'm will be very happy.
BTW, the list of modules I would like to contribute ( _after_ 3.3
development is over):
- Servlet2.3 facade
- reuse the log valve
- pick up and use the JNDI abstractions
- reuse the HTTP1.1 interceptor
- merge mod_webapp with mod_jk
- improve mod_jk performance
Some of those will probably be done by others, but I'll try to find time
to help them. I realize I will not be able to do all of that, but I am
confident other people will help. If you don't like that - too bad, but I
don't think your vote will change that - it'll just mean that apache will
loose good contributions.
I think it'll have a nice evolution ( with the inconvenience that users
will have to download the modules from a different place). As I said, if
Apache is not wanting to see those or doesn't want 3.3 to have any future
- too bad.
Once again - I'm not going to fork, and I'm not going to abandon 3.3 as
long as there are users using it. I'll just do it in a different place.
Costin
