[Mlview-hacking] Hello ...

[Frederick C. Druseikis]

Just wanted to say hello.

I've been working on an XSLT-based web application for some time now.  I've
used mlview for a couple of *days* mostly because I started feeling the need
for an XML editor.  I have been pretty impressed with mlview.  It hasn't
bitten me yet ;) Although I've been using vim to edit my .xsl I'm beginning
to think that I could really use a more sophisticated set of editing tools,
so that's why I joined this list.

The approach I've been taking is that there are meta-stylesheets which when
coupled with metadata (describing page content) produce a stylesheet that
takes data to produce an html page.  Metastyles sheets get really interesting
because there is a lot of "quoting" of xslt constructs to get them to produce
style-sheets (lots of xsl:element and xsl:attribute constructs).  Mlview
has a lot of knowledge via a dtd of an xml document; constructs for xslt
have a lot of structure too.  Has anybody tried to use mlview for this kind
of task?  Has anybody seen (or have) a dtd for XSLT?

Does anybody have any interest in - or percieve a need for - adding a "plug-in"
capability to mlview?  I'm thinking that a capability like that found in
gedit.

What I have in mind is that you would be able to write a plug-in (say in
perl, or python, although I prefer the former) that would operate on the
DOM mlview has, maybe by taking a copy of the current one, passing it to
the plug-in, and then displaying the result on a new tab.  Pretty clear that
it would need to use the libxml2 library that mlview is using, which is true
for perl.  What I'd like to do with it is to add tools that allow you to
do quoting of XSLT constructs, creating xsl:templates, parameterize existing
xslt code, etc..  A possibility is to do it all in XSLT, althought that seems
slightly too academic to me.  Some of these kinds of things seem like simple
recursive algorthms in a tool like perl operating on a DOM.  So I'm thinking
that separating the pure viewer tasks from the transform tasks is a good
way to go.  I haven't thought it out beyond this.

Regards,
Fred

--

Frederick C. Druseikis, Ph.D.
Adjunct Professor - Department of Computer Science and Engineering
Senior Research Fellow - Center for Health Services & Policy Research
Swearingen Rm 1A01C
University of South Carolina
Columbia, South Carolina  29208
address@hidden
http://cse.sc.edu/~fredd