errata in general and on 'language' datatype

Dear XML Schema errata maintainers,

This mail contains two points.

I just checked XML Schema Part 2 errate with respect to
the 'language' datatype. I found
http://www.w3.org/2001/05/xmlschema-errata#e2-25,
which is great. However, this took me some time.
In particular, I had to do 'view source' to find out
that you used the error numbers as fragment identifiers
(which I guessed correctly), but with a lowercase 'e'
instead of an uppercase 'E'. These fragment identifiers
are case-sensitive (as they were since the days of the
first HTML implementations). Would you mind please to
change the labels to lower cases (changing the fragment
identifiers is a bad idea because that will make many
links go south).


Second, erratum E41 to XML 1.0 second edition
(http://www.w3.org/XML/xml-V10-2e-errata#E41) adds the
possibility to have xml:lang="". This is important for
various scenarios ranging from SOAP to RDF to XML Singatures.
This is currently not allowed for the 'language' simple type.
While strictly speaking, this is an unforseeable change,
I'm wondering how it the current version of XML Schema
the absence of language is indicated (an absence which
always was possible even before this erratum to XML 1.0,
at least at the top of the tree and as far down as the
first actual xml:lang. Would something like xml:lang="xsi:nil"
have done the job?

This erratum also changes DTD declarations for xml:lang
from NMTOKEN to CDATA; fortunately XML Schema does not
make xml:lang a subtype of NMTOKEN, so there is no change
in the type hierarchy needed.

As to whether it is appropriate to specify a pattern
for this type, see also
http://www.w3.org/XML/xml-19980210-errata#E73.


With kind regards,    Martin.

Received on Monday, 14 July 2003 11:27:23 UTC