> -----Original Message-----
> From: plug-discuss-admin@lists.PLUG.phoenix.az.us
> [mailto:plug-discuss-admin@lists.PLUG.phoenix.az.us]On Behalf Of Shawn
> T. Rutledge
> Sent: Friday, August 10, 2001 2:35 PM
> To: plug-discuss@lists.PLUG.phoenix.az.us
> Subject: Re: Advice Please: SAXON, Xalan, JADE, LaTeX, and XSL:FO
>
>
> On Mon, Aug 06, 2001 at 02:13:13PM -0700, Trent Shipley wrote:
> > 30: SAXON vs Xalan vs JADE
> >

>
> See also this:
>
> http://www.xml.com/pub/a/1999/05/xsl/xslconsidered_1.html
>
> I think he's arguing to use DSSSL for hardcopy documents (because it's
> more powerful) and CSS for electronic documents (because it's more suited
> to electronic documents, easier, and preserves the XML's meta-information
> better).  But I don't have enough experience with all of these to be sure
> whether he is right.

XSL Considered Harmful
by Michael Leventhal
May 20, 1999


Leventhal wrote about XSL before it was split into XSLT and XSL:FO

He argues:

1) CSS-2 + DOM are entirely adequate for interactive web applications.

2) XSL:FO can't be used for interactive documents.  It is for printed stuff.
The W3C shouldn't be concerned with printed stuff.  The W3C is therefore
wasting its time on FO and I am really angry at them because they could be
doing more important stuff.

3) DSSSL is so hard that even Michael Leventhal can't use it.  It is really,
really hard.  XSLT is just DSSSL repackaged.  It is just as hard and way too
weak.  Do all your document transformation using a generic EZ-to-write 3GL
language like BASIC, JavaScript, or PERL.  (??!)
     In _theory_ you could use XSLT for some part of developing or deploying
interactive content, but nobody is going to.  The W3C is wasting its time
here too.  My goodness, I am so very angry at them.

4) You can't use regular expressions in XSLT (or presumably DSSSL).  This is
superceeded by the X-Pointer standard.  Also, a properly implemented DSSSL
compiler should be built over Scheme, in principle this should give access
to a RegEx library.

-----

Obviously item 2 is irrelevant to me precisely because I intend to output my
dissertation as a printed document.

Item three I find hard to fathom.  DSSSL is doesn't look that hard to use to
me.  Recombining objects into procedural code every time you want to
re-order or alter the layout of a primary set of source text seems just
hideous.

DSSSL/XSLT may not be the 4GL declarative solution, but there must be a way
to engineer a usable 4GL declarative language for markup.


> If all else fails you could bypass TeX and write directly in
> Postscript, or
> write something to convert your SGML to Postscript.  Maybe you
> could try the
> "gonzo" package at www.tinaja.com and modify as necessary; it has basic
> routines for page-layout.  But I haven't used it myself yet.  I have
> written only very simple Postscript (for example to print envelopes on
> my laser via Linux - I couldn't find any other way to do it).  I added
> a feature to Nettebook for this; I start with a Postscript "template"
> which has XML tags in it; I replace the XML tags with appropriate stuff
> from the database (name, address etc.); and then send it to lpr.  The
> result is I can select a person in the database (via the web interface)
> and print an envelope.  The tag-substitution is done via simple find-and-
> replace in PHP.