Re: How could SGML help? (was RE: Blueberry/Unicode/XML)

[John Cowan <jcowan@reutershealth.com>]

Mike.Champion@SoftwareAG-USA.com wrote:

> I for one don't want to flame or be flamed ... just to better
> understand what the SGML solution for the "Blueberry" problem would
> be and if there is anything in SGML that could plausibly be incorporated
> into "XML 1.1" to solve the problem in a more generic way. 

SGML has the concept of the SGML Declaration.  A document can point

explicitly to the SGML Declaration associated with it, or an SGML
parser can have a default declaration.


Among other things, this enables you to declare exactly which
characters are used for start-of-tag, end-of-tag, attribute-equals,
and every other syntax role.  In particular, you can declare which
characters can appear in names, either in any position, or in any
but the first.

The default default SGML declaration, used if the parser has no
other clue, basically specifies the XML syntax with a few exceptions.
The default at the end of a PI is just ">" rather than "?>", and
only ASCII characters are allowed in names.

The designers of XML decided to sacrifice all this seldom-used
flexibility in favor of a single syntax, spelled out in the
XML Recommendation.  In order to change the syntax, therefore,
we have to change the definition of XML.

That is what the Blueberry effort is looking to do, in a
safe and backward-compatible way.

-- 
There is / one art             || John Cowan <jcowan@reutershealth.com>
no more / no less              || http://www.reutershealth.com
to do / all things             || http://www.ccil.org/~cowan
with art- / lessness           \\ -- Piet Hein