Re: XML Blueberry (non-ASCII name characters in Japan)

[David Brownell <david-b@pacbell.net>]

A few years back, I noticed that most XML parsers had logic
that went pretty much like:

    if (!"1.0".equals (version))
        panic ()

It might be an appropriate time for folk to start planning how
that logic will need to change.  Perhaps

    if (!("1.0".equals (version) || "blueberry".equals (version) ))
        panic ();

will be the way to go?  I don't really like the idea of accepting
just any version string there, because of a suspicion (grounded
in history) that all kinds of XML "versions" will start to show
up, perhaps starting with "VendorM" (or "VendorI", etc) with
all sorts of exotic rules about what each variant means (this
one just deletes DTDs, that one embeds W3C's schemas,
this other extended the standard with proprietary syntax).

Note that the issue isn't "what should the string be"; it's "how
should such a transition be handled", along with "what should
the interoperability consequences be".

- Dave