Re: (dsssl) Heresy

[Adam Turoff <ziggy@xxxxxxxxx>]

On Fri, Apr 04, 2003 at 09:02:49PM -0600, Paul Tyson wrote:
> In the last couple of years I have chosen to use XSLT because I 
> didn't want to saddle my clients with the superior, but more 
> abstruse DSSSL technology.  I am now convinced that only by 
> severely limiting your expectations will you be satisfied with 
> XSLT.  I made the mistake of not scaling back my imagination to 
> suit the technology.

I was in a meeting this week with a client discussing the care and
maintenance of their DSSSL-based system.  At the end of the meeting, the
only benefit we found to an XSLT migration was the possibility of
finding more potential employees/consultants/agencies that have XSLT
experience, or the increased ease of teaching existing employees XSLT
vs. DSSSL.

A year ago, I talked with this client about converting their
SGML-based document formats into XML, and moving away from DSSSL
to XSLT, because they document styling they want to perform is
theoretically possible in both.  (They produce print formats
exclusively.)

First, the XMLification of their document format faultered, because
there are portions of the system that are raw HTML fragments.  Each
document requires another few special case hacks to accommodate
malformed XML fragments that are still semi-valid SGML.

Second, we learned that there was a very important business case
that required RTF output.  Make all the noise you want about how
PDF is a much more precise and controlled output representation,
but there were customers that absolutely needed RTF, and a lot of
documents where briefly hand-editing a "final" document was
significantly more cost effective than spending hours or days to fix a
broken document/program.

Finally, I was asked recently if more cost effective to develop DSSSL or
XSLT going forward.  After a few moments thought, I realized that it
would cost *more* for this client to manage a body of XSLT stylesheets.
(Keep in mind that there is a large body of pretty nasty looking DSSSL
to draw upon and improve, and no existing XSLT infrastructure
whatsoever.)  Even if there were an XML->RTF path available *right* *now*,
it simply takes (me) longer to code, test and debug programmatic
transformations with XSLT than it does with Scheme.

Just $0.02,

Z.


 DSSSList info and archive:  http://www.mulberrytech.com/dsssl/dssslist