Hi Rick and other guys!
First of all, thank'x for this very impressive servlet api implementation.
it's really good and promisses!
I was investigating alternatives to jasper on tomcat when I found wistone.
Then I found one message at tomcat-devel where Rick asked some thinks about
jasper and BCEL.
I'm thinking about this and I believe that one could replace de
jdt-compilation in jasper or even write a new jsp implementation using
byte-code generation strategies...
What you've found Rick? Anyone else attempting to implement this?
I'm not the very best HTTP/Servlet programmer, but I want to help and learn=
PS.: Excuse my bad English, I'm brazilian ;)
Caetano Spuldaro Neto
Joa=E7aba, SC, Brasil
From: Rick Knowles <rick@kn...> - 2006-07-22 12:03:10
Caetano Spuldaro Neto wrote:
> Hi Rick and other guys!
> First of all, thank'x for this very impressive servlet api
> implementation. it's really good and promisses!
Thanks ... glad to hear it's useful.
> I was investigating alternatives to jasper on tomcat when I found
> winstone. Then I found one message at tomcat-devel where Rick asked
> some things about jasper and BCEL.
> I'm thinking about this and I believe that one could replace the
> jdt-compilation in jasper or even write a new jsp implementation using
> byte-code generation strategies...
> What you've found Rick? Anyone else attempting to implement this?
There was actually a follow up discussion with Remy Maucherat (the lead
developer of Tomcat) on that same mailing list you saw. He raised a very
important point that I didn't notice. When I suggested BCEL, it was
because my own JSP programming was almost completely focussed on tag
library based designs (ie the JSP page doesn't contain any java syntax),
and if that is all JSP required, BCEL and other byte code generation
techniques would be much faster than JDT and javac.
Remy's point was that the JSP specification requires "Scriptlet" support
(ie embedded java code in the JSP page). I never use this, so I didn't
notice it. The overhead cost of parsing this scriptlet java syntax would
make any speed advantage of using BCEL disappear. Also, to implement
this properly would require using a parser generator, which I know very
little about, but I've heard they don't produce small compact code.
I guess I wouldn't be against it, but I've noticed that most people I
know use XSL (or some other templating language) rather than JSP anyway,
so I don't really have enough motivation to do it myself. I'm happy to
help you with integrating it into winstone if you like, but I can't
really justify the time required to write a JSP implementation with you
Sorry to be so negative ... that was just where my thinking about this
topic ended before. If there was something that we could make that was
significantly different and valuable I'd be in it, but this particular
idea fails because of the small details.
> PS.: Excuse my bad English, I'm brazilian ;)
Not at all ... your English sounds fine to me. At any rate, it's much
better than my Portuguese :)