[
Lists Home |
Date Index |
Thread Index
]
On Apr 9, 2004, at 11:32 AM, Elliotte Rusty Harold wrote:
>
> There are some use cases where XML is just not appropriate. I
Right, but "XML" means many things to different people. There are lots
of people out there who have a use for XSLT, XPath, XQuery,
DOM/JDOM/whatever, and other technologies that are based on some
variation of the XML Infoset rather than syntax .... but the XML 1.0
syntax is clearly inappropriate. What are they to do? They are in the
fuzzy zone where some aspects of XML are appropriate and others are
not, so it's unrealistic to expect them to forego the real benefits
they do get from this stuff [Sigh, this is a permathread, sorry ]
> . However, for what it does work for, it works very well; and trying
> to make it work better for some of the current use cases, or for new
> use cases, at the expense of existing use cases and interoperability
> does not seem wise.
>
I've yet to hear of a proposal that would come at the expense of
interop for mainstream XML-syntax use cases. The alternative for most
people in the situations I discuss is not "bite the bullet, buy more
hardware, and use XML syntax" -- it's to do something proprietary and
*intrinsically* non-interoperable. The question -- a good question,
and I don't have a position on it -- is whether there is any value in
coming up with one or more standardized alternative syntaxes with which
to serialize the XML Infoset (or XQuery data model, or whatever) as
opposed to letting every application or vendor define their own.
Prematurely standardizing a "binary" format optimized for one scenario
could indeed create more pain than gain, but that was the consensus
that came out of the W3C workshop and goes into the Binary
Characterization WG. The argument that we must not even identify the
real bottlenecks or explore alternatives because that would somehow
undermine interoperability in the (1997-vintage) mainstream use cases
just doesn't make sense to me.
|
