Re: Efficient XML - request for input

["Michael Champion" <mike.champion@softwareag-usa.com>]

----- Original Message -----
From: "Matthew MacKenzie" <matt@xmlglobal.com>
To: <donpark@docuverse.com>; "'Duane Nickull'" <duane@xmlglobal.com>;
"'ebXML-Transport@lists. oasis-open. org'"
<ebXML-Transport@lists.oasis-open.org>; "'ebXML-Architecture List'"
<ebXML-Architecture@lists.oasis-open.org>
Sent: Thursday, March 09, 2000 11:40 AM
Subject: RE: Efficient XML - request for input


>
> Okay, I don't know where I would be without my CDATA :), but there is a
> good lesson to learn from SML - keep the markup as clear and simple as
> possible.  Every time I see something like XMI or RDF I cringe, because it
> is just too complex for what it is being used for.

Right.  (BTW, something like "Minimal XML" or "XML Lite" is probably a
better term than SML, since it's used for so many other acronyms).  The
point is to start with the bare minimum of XML syntax, and add more only as
really needed.  As much as I detest CDATA sections in the abstract, they are
very useful for embedded script (like XMetaL's macro files).

The XML world sortof reminds me of the C++ world 10 years ago -- people seem
to like to play with the most obscure features just because cool people use
them ;~)  Namespaces, RDF, etc. have their place, but I would encourage
ebXML to start very simple and add complexity (even attributes and mixed
content!) only when there is a compelling need.  Rather than, for example,
assume that namespaces are obviously needed and that RDF is *the* way to
describe metadata, employ them ONLY when they provide the necessary bang for
the byte (or cycle, or learning curve, etc.).