Lists Home |
Date Index |
firstname.lastname@example.org (Michael Rys) writes:
>It is this narrow approach to split data into different categories that
>XML transcends. Together with the simple and standardized (single)
>syntax, this simplifies interoperability of documents AND data.
>Oh well, I know that I will never get Alaric to see beyond this narrow
>markup text -> XML, data -> something else (like ASN1) division.
Hmmm... perhaps you swallowed the XML hype a little too completely?
While I don't share Alaric's motivations particularly, I think he's
quite right that XML is not necessarily the right solution for data,
'transcendent' claims not withstanding.
The current state of XML's 'transcendence' is actually quite depressing
to me. XML has won the tools battle, in that XML parsers and processors
are more widely and cheaply available than anything else similarly
generic, but XML appears to be utterly losing the battle if you look at
how people use XML, whether they care that they're using XML, etc.
There's a major culture clash between markup approaches and pretty much
every other form of data handling, storage, and interchange in the
computing world. The choices markup makes - human-readable data that
can be manipulated directly by users - are very different from the
priorities of the rest of the data world, which generally insists on
applications as intermediaries between humans and data.
The different choices come with different costs and benefits. XML is
verbose, not particularly amenable to graph-oriented visions, and syntax
has consequences. Other approaches typically focus on compactness, rely
on graphs rather than trees to represent data, and bad data has
consequences but syntax is supposed to be inaccessible anyway.
A lot of the XML vocabularies I see - definitely including those from
Michael's employer - appear to have no understanding of this difference.
Object structures get serialized into XML for the sake of XML, with
little or no attention paid to what would actually make it look good in
XML or be most useful with the XML toolkit. Because people have
swallowed this "transcendent" Kool-Aid, they just take ideas from
programming media and dump them into markup media without concern for
the consequences for either their data or XML craft.
It's a sad cycle - one I'd really like to see technologies like ASN.1
disrupt. XML is great for some things, lousy for others, and pretending
otherwise just piles more and more useless crap into XML. W3C XML
Schema is examplar #1 of that, but the problems are deep, and not
limited to WXS.
The ubiquity of XML is now very impressive. The quality of that
ubiquitous XML is pretty damn horrible. The low quality of that XML
drastically reduces the long-term benefit of using XML, even if it seems
cheap and will do for now.