> If you specify the full DTD in the external subset and then redeclare IDs
> in the internal subset some parsers will warn about duplicate declarations
> (its only a warning but standardising on something that parsers warn
> about seems odd)
... to me it's seemed odd to generate warnings _by default_ about
constructs that by definition are no problem. Warn if the second
(ignored) declaration conflicts with the first one, sure -- there's a
chance that might actually indicate a problem.
> None of these problems is impossible to work round, but surely it is
> legitimate to ask if it's possible to set up the ground rules in a way
> that makes all of this easier to describe?
That might be the main reason to want "xml:id" or somesuch, rather than
copying ID attribute declarations into the internal subset. But let's not
forget that the internal subset solution _does_ work except in the case
of folk who've adopted anti-DTD policies ... :)
- Dave
