Re: Future XMLers (was document management and xml)

["Mark Baker" <mbaker -at- omnimark -dot- com>]

garret.h.romaine wrote



>XML seems like the way to go; it's a standard whose time has come. SGML
>looked too complicated and unwieldy,

I don't want to seem to be raining on XML's parade, but, just so that
unrealistic expectations don't lead to disappointment, I will point out that
the only way in which SGML is more complicated than XML is if you are a
parser writer. (The parser is the piece of software that reads a tagged
document and figures out which bits are tags and which bits are document.)

Both XML and SGML are meta-languages used to write descriptions of tagging
languages so that those tagging languages can be read by standardized
parsers. I doubt anyone on this list is ever going to write a parser. We are
going to design and use tagging languages.

The only reason technical writers should be interested in XML is if they
want to create custom tagging languages that will let them express the
structure of their data in a way not supported by any existing data formats.
Our challenge will be to design effective tagging languages that capture the
data we need and are easy to use for authors.

This is a difficult thing to do. It is not any easier to do with XML than
with SGML. In fact, it really has nothing to do with XML or SGML per se. It
is a problem in semantics, in describing and assigning meaning to units of
information. XML just provide a possible syntax for expressing those
semantics.

My point? The question you should ask is not "Do I use XML or don't I?", but
"Do I create custom data structures or don't I?" Creating custom data
structures is hard. Using XML is easy, but irrelevant unless you need to
create custom data structures.

(BTW, on a purely techie level, I would have to say that since the namespace
spec got approved, XML is now more complicated and unwieldy than SGML ever
was at the syntactic level. Fortunately you can and should ignore all the
complicate bits of both XML and SGML.)