Re: My opinion on the problem of identifying ID semantics

> Can you provide more detail on both the lack of power and the
> increased cost?

I would like to use multi-part keys.  For example, the attribute
"family" and the attribute "given" collectively identify a person.

The xml:idAttr attribute sketched in the Draft TAG finding document
cannot represent multi-part keys.  Furthermore, it cannot use integers
as keys; that is, "1" and "01" are assumed to be different.  I thus think that
xml:idAtt is a short-term solution.

Although the xml:idAttr attribute is not a significant burden, I do not
want to have another mechanism that may eventually tie our hands when
we develop a long-term solution.  Our hands are already tied by
key/keyref/unique of WXS and ID/IDEREF/IDREFS of DTDs.

Cheers,

-- 
MURATA Makoto (FAMILY Given) <EB2M-MRT@asahi-net.or.jp>