On 20 Oct 2001, Peter Mutsaers wrote:
>
> Mangi> rather use a word processor. If we push the tomcat engine
> Mangi> towards being a fully featured webserver isn't it
> Mangi> eventually going to bloat? Won't the task we asked it to do
> Mangi> (handle servlet/jsp requests) wind up suffering?
>
> The 'bloat' in fully featured webservers doesn't come from being a
> webserver itself (which is trivial); rather it comes from all kinds of
> authentification schemes (which in a normal Servlet implementation
> you'd solve in Java anyways, the JDBC realm being an example, or the
> JNS (java naming service) features in tomcat. Or it comes from having
> modules for all kinds of scripting languages (php, cgi, etc etc) that
> are irrelevant in a non-legacy servlet environment.
>
I would only be concerned about "bloat" if I was forced to suffer
performance hits for all the stuff that *I* am not using.
Consider Apache itself -- it ships with around 100 available plug-in
modules that can do all sorts of wierd and wonderful things, and there are
lots more available around the web. But you only plug in the ones you
need. You get what you pay for.
Many of the "extra features" in Tomcat 4 already work that way. You only
have to worry about them if you use them. For example, container-managed
security is implemented by using a plug-in Valve that is installed ONLY
for web apps that use it. No other webapp is affected at all. Likewise,
access logging, CGI, SSI, and many other things are modular components
that you can assemble to suit your needs.
In such an environment, a webapp that needs just the basic servlet/JSP (or
even just servlet) capabilities does not suffer. Yet people who need full
featured capabilities can still use the same basic platform. To me,
that's the best of both worlds.
Craig
