Lists Home |
Date Index |
- From: "Rick Jelliffe" <firstname.lastname@example.org>
- To: "Jonathan Robie" <email@example.com>
- Date: Tue, 30 Sep 1997 04:25:53 +1000
> From: Jonathan Robie <firstname.lastname@example.org>
> Of course, all existing SGML and XML tools know how to deal with DTDs,
> and this is a rather major departure from traditional SGML. It has not
> been blessed by any standardization committee. Given the way Microsoft
> has approached Java, insisting that it need not implement the portable
> libraries everyone else is using, and encouraging people to use their
> platform-specific libraries instead, it is easy to wonder what will
> happen to the SGML world if Microsoft is in control of an alternative
> method of specifying content models.
XML-data would probably fail, that's what.
Because their form of schemas are so complicated and verbose to read
that you will need browsing tools to manipulate them. This in turn
gives schemas (even though they are written in XML) the nature
of binary objects rather than textual objects. It seems the weight
of experience is against people making successful schema languages
in non-textual forms.
For example, Bento and the OpenDoc storage system included API-driven
routines for decorating cleverly stored objects with all sorts of
interesting type information, including type conversion, and so it
can be considered -- in part -- a schema system. Failing to
have a text form, the thing failed to thrive. The XML-data
system does have a text form, but it complicates matters so much by
not having a simple text form (e.g. a separate declaration
syntax) that it seems to be unreadable.
In my view, declarations are actually a kind of processing instruction,
targetted at the parser or entity manager, which also may be of
interest to the application (sorry for using SGML jargon).
The XML-data view seems to be that they are, more essentially,
data rather than processing instructions. Tim Bray has said
frequently "metadata is data", to which I would say
"processing instructions are sometimes data, sometimes not".
Have the XML-data people ever made any requests to ISO for
suggested improvements to the declaration syntax to give
them the functionality they need? (This is unfair really,
since I think XML-data is an experimental system, and
therefore a good place to generate user requirements for
a less verbose syntax.) Have they proved that
a single-tag language is easier to use than one with multiple
types of tags?
I am certainly in 100% favour of schema systems and stronger typing
and abstracting interesting information about data into
header elements. I proposed the SEEALSO parameter in the
current WebSGML TC specifically to allow richer declarations
of syntax using any kind of exotic notations including natural
language, so I am the last person to say that SGML declarations
are enough for all uses.
But I am simply not convinced that XML-data represents a
usable alternative to the standard declarations (in the
same market), and I think XML-data should not compete
(or been talked about as competing!) with the standard
declarations. Their purposes are, I hope, quite
xml-dev: A list for W3C XML Developers. To post, mailto:email@example.com
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/
To (un)subscribe, mailto:firstname.lastname@example.org the following message;
To subscribe to the digests, mailto:email@example.com the following message;
List coordinator, Henry Rzepa (mailto:firstname.lastname@example.org)