HTML-MARC

[Marc Salomon]

|What about the character set? Or is it implied in one of your categories?
|In Finland we use FINMARC, but I think there are variations in the
|character sets between different user bodies.

HTTP already has an Accept-Charset: header to specify desired charsets, but it
might be nice to specify an alternate charset in cases when MARC was served by
protocols with weaker negotiation facilities.

|But there are so many different MARCs...Understanding the semantics of
|each and every one could be quite a task even if there weren't any errors
|in the records...Good luck anyway.

That's why negotiation is used to limit the set to the comprehensible.

|But, it would make sense if the metadata had to be in the same file as
|the electronic document provided that the document would also be encoded
|in SGML. And the size limit of a marc record is 100KB (admittedly it is
|sufficient for most needs but there is no size limit in an SGML instance).

The only reason to keep metadata in the file it describes is either a very
small, managable collection or laziness.  I don't think that a DTD can capture
the rules of MARC encoding, especially when it comes to rules like "You can't
have a x subfield unless you have a w subfield..."

|I have a SGML DTD for USMARC - I remember having downloaded it from
|Berkeley...So anyone who is doing such a job could think twice whether
|re-inventing the wheel is really necessary.

Yes, I was referring to Ray Larson's DTD for Cheshire.

-marc

--