---- Included message:
- Does not allow at all the use of Norwegian in tag names.
(for instance, the Norwegian word for "birth" contains an O-slash).
The concern for use of international character sets in name strings
was a moderately hot topic during the Mime discussions and it has
been mentioned a number of times about STIF.
We have no quarrel about the general need to support international
character sets and languages, so the only question is when and how.
One model could be: Always. Any string may be in any character set.
Another model could be: All "values" may be in any character set,
but the "tags" for the values are more restricted.
Other forms for the model are possible, of course, but I find myself
favoring this second framework.
Standardizing a framework for doing attribute/value exchanges requires
that there be some common base. For the Internet, that base must work
for the whole Internet, in my opinion. To the extent that some
information is special to a certain community, I believe we MUST
allow that community a means of expressing the information. To the
extent that the exchange is intended for the entire Internet, then
it MUST be universal. STIF is intended for exchanges on the Internet.
It is attempting to provide a base for standardizing sets of attributes.
This MUST include the NAMES of those attributes, since that is the only
way to refer to them. Allowing variation in charactersets for that
part of the specifications will, I believe, introduce optional behavior
that only adds to complexity and does not faciliate interoperability.
This is one point of view. I'd be interested in hearing counter-views.