Re: short-tag considered unhealthy
James's design criterion sounds like a good one. Empty end-tags and
unquoted attribute values do indeed fall into the class of things I had
looked at and said "It's not too hard, let's leave it in", but which are
also clearly inessential. Any doubt I had in my mind went away when I
asked myself "Well, if this feature isn't in XML, will you be forced to
write a pre-processor for it, to translate your working format into
There are certainly things I can imagine writing preprocessors for, such
as multiple attlist or element declarations, conditional sections, or
other things the TEI DTDs currently do with marked sections and
parameter entities. And my sense that others besides the TEI need such
functionality might be a reason for including that function in XML.
Or maybe not: maybe some things we need can be embodied in such
preprocessors, rather than built into the core of XML.
But on empty end-tags and unquoted attribute values, I think James
has persuaded me. They are useful for minimizing keystrokes when using
dumb software; they are *not* necessary or even particularly useful
in providing better data structures.
-C. M. Sperberg-McQueen