Return-Path: Mailing-List: contact tomcat-dev-help@jakarta.apache.org; run by ezmlm Delivered-To: mailing list tomcat-dev@jakarta.apache.org Received: (qmail 27944 invoked from network); 20 Oct 1999 02:57:10 -0000 Received: from multit.lnk.telstra.net (HELO gateway.multitask.com.au) (139.130.63.146) by apache.org with SMTP; 20 Oct 1999 02:57:10 -0000 Subject: multiple languages... To: tomcat-dev@jakarta.apache.org Cc: Bcc: X-Mailer: Lotus Notes Release 5.0a (Intl) 4 May 1999 Date: Wed, 20 Oct 1999 02:56:39 GMT Message-ID: From: dion@multitask.com.au X-Priority: 3 (Normal) X-MIMETrack: Serialize by Router on gateway/Multitask Consulting/AU(Release 5.0a (Intl)|4 May 1999) at 20/10/1999 13:00:54 MIME-Version: 1.0 Content-type: text/plain; charset=us-ascii "We did not pursue other languages initially due to several reasons, including that they complicate the portability story, but if there is real interest, I'll be very happy to help with the effort. I personally find that I am writing more and more tag libraries and not using as many scripting elements (also using more nesting for action communication, rather than scripting variables). Note that a tag library essentially defines a specialized language." The only problem I see is that writing a tag lib is a much higher entry learning curve for a newbie. "I am not convinced that we can easily mix-and-match multiple scripting languages within a page. The problem is that each language has its own syntax and semantics, and the scriptlets in JSP are not even fully formed statements, they can be anything (as long as... see the spec for details), so defining what is a well-formed page in the presence of multiple program fragments of multiple languages is going to be "interesting". Specially without imposing unwanted constraints on the implementation. One can probably do better by restricting what can be used in the scriptlets, but note that some people really like to say things like: <% for (int i=0; i<34; i++) { %> stuff.... <% } %> or even more convoluted things... But lets' see...." I'm with you on this one....for example doing something like: <% if (cond) { %>

Blah <% } else { %>

Blah blah <% $perlcode = $cond; %> <% } %> is theoretically possible, but not a good option IMHO. I'm also interested academically in making the DOM available to javascript *during* JSP processing, along with the implicit document object, sans window.... "Going through an XML rep seems reasonable, although we will need to coordinate it with the JSP.next effort when that gets moving (see above). I would not worry too much about things like the att="<%= ...%>" stuff in the current syntax; we have a lot of freedom in the XML mapping - the JSP 1.1 and JSP 1.0 specs explicitly (and for this very reason) indicate that the mapping indicated there may change. And yes, we want to use XSL and to feed to XSL (and other transformations). Some of the latter can already be done in JSP 1.1 although nowhere as nicely and/or efficiently as I want." Currently the code does most of the code generation work at end of document processing, so it theoretically isn't too hard to change the existing implementation to build an XML document on the fly during the handle* methods and then process at the end...is this what you'd envisioned?? "Hope this helps, It is great to see all this interest and energy. I think we are headed for a few very "fun" months." Other things in this context on my 'hot list' are command line compiles for example....is that currently done? I couldn't find a main() in org.apache.jasper.compiler.JspCompiler... -- dIon Gillard, Multitask Consulting Work: http://www.multitask.com.au Play: http://www.softwaresummit.com I'm presenting at the Java Programming Conference "Colorado Software Summit" this year, see you there!