tomcat-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Craig R. McClanahan" <Craig.McClana...@eng.sun.com>
Subject Re: [PROPOSAL] Tomcat 4.0 Milestone 5
Date Fri, 24 Nov 2000 20:00:06 GMT
See intermixed below.

GOMEZ Henri wrote:

> >I plan to cut a "milestone 5" release of Tomcat 4.0 this coming Friday (or,
> >worst case, on Saturday).  As a sneak preview of coming attractions, the
> >following cool stuff will be included:
>
> Nice...
>
> >* Initial version of the web connector for Apache
>
> - mod_warp ? Will you include documentation about it ?
>   Will it support load-balancing and fault-tolerance à la mod_jserv/mod_jk ?
>   I'd like to see if we could use it in 3.2 branch.
>
> >After milestone 5, I propose that we consider Tomcat 4.0 to be "feature
> >complete", so that we can commence with a beta cycle, and focus on shaking
> out
> >remaining bugs -- aiming for a final release either late December or early
> >January.
>
> Your planning is to release TC 4.0 for late December or early January ?-)
> What about Tomcat 3.2 release ? early December ?
>

There have been several bug fixes posted for 3.2b8 that I want to evaluate, but
no new critical bugs.  Therefore, my inclination will be to cut a final 3.2
release next week, and start doing a maintenance cycle (3.2.1, 3.2.2, etc.) as
needed -- we should discuss this more next week, after all the US folks get over
their turkey hangovers ;-).

>
> Excuse me but I saw you there a little too optimist.
>
>

On 4.0 stability?  I don't think so.  You should consider me to be biased
(Catalina has been *my* baby, after all :-), but I've seen the insides of 3.2
and 4.0 in a lot more detail than I ever hoped to over the last month.  I have
come to the following conclusions:

* The servlet container code in 3.2 is *substantially* better
  than the servlet container code in 3.1 in terms of spec conformance,
  number of bugs, functionality, and performance.

* The servlet container code in 4.0 is "more better" than 3.2
  on spec compliance, at least as good on bug counts, and
  at least as good on performance (soon to be better, because
  Remy is working on optimizations on the HTTP/1.1 stack).

* The servlet container in 4.0 sports many additional features
  over 3.2, and only lacks one at the moment -- the ability to
  run a web app under a security manager.  That feature can be
  added in 4.1, as soon as someone who understands this stuff
  (Glenn, hint hint :-) can help make it happen.

* The JSP code in 3.2 and 4.0 is pretty equivalent, because we've
  been tracking the bug fixes pretty well in parallel.

There are (at least) two reasons where you would HAVE to choose one version or
the other:

* Do you need web connectors?  Go with 3.2 -- this area is
  obviously not mature yet in 4.0.

* Do you need servlet 2.3 or JSP 1.2 features.  Go with 4.0
  -- obviously, that is the only place these are implemented.

For any other scenario, you owe it to yourself to evaluate both platforms.

Just as a point of interest, Tomcat 4.0 milestone 4 has been downloaded just
under 9,100 times since it was posted on November 1.  There have certainly been
some bug reports, but not a large volume -- given the download counts, this
seems to be due more to code quality than people ignoring it :-).

Just as another point of interest, every time I posted a fix to the 3.2 tree
over the last couple of weeks, I also looked at whether 4.0 suffered from the
same, or a similar, problem.  On the JSP side the bugs were pretty consistent in
both (natural enough, they were the same code base).  On the servlet side, it's
happened twice out of about 35 bug fixes -- in both cases where I had copied
utility code directly from the 3.2 code base, and therefore carried forward
bugs.

> >At the same time, we can start talking about features and
> >functionality we'd like to see in Tomcat 4.1.
>
> Whaoo, Tomcat 4.1. Allready ?

Yes.  From my perspective, we should consider 4.0 to be in feature freeze with
the addition of the web connectors, so future functionality additions should be
planned for a later release.

>
> But what about support for Tomcat 3.2.1, 3.2.2, 3.2.3.... (bug fixes)
>
> It seems clear for you that Tomcat 3.2 (and the full 3.x line)
> are allready products in end of life.
>

I don't think that is actually the case.

Large numbers of people have been eagerly awaiting a 3.2 final release.  Many
tool and other vendors have integrated Tomcat 3.2 betas into their products, and
it is not appropriate to encourage them to drop it.

I will be happy to continue applying contributed patches to the 3.2 code base,
and releasing 3.2.1, 3.2.2 type releases as necessary -- pretty much along the
same line that Apache JServ has continued to create bug fix releases as needed.
I'm not planning on doing any new development on the 3.2 track, however -- the
future path is obviously 4.0.

>
> But did they even have the time to start their 'production life' ?
>
> I'd like to see what the others contributers of Tomcat thinks about
> that.
>
> I'm sorry to say but THERE IS STILL NO STABLE TOMCAT RELEASE.
>

Aside from the fact that "beta" doesn't really mean the same thing in open
source as it does in commercial products, your point is well taken.  That's the
primary reason I went back and picked up the pieces of 3.2 that everyone else
abandoned.  The wait for 3.2 is almost over.

>
> Home page : http://jakarta.apache.org/
>
> 'The goal of the Jakarta Project is to provide commercial-quality server
> solutions
> based on the Java Platform that are developed in an open and cooperative
> fashion'

Craig



Mime
View raw message