Re: XSV - xmlns errors

On Fri, 2002-02-08 at 11:32, Donna Bergmark wrote:
> We are still happily using XSV to validate potention OAI repositories
> against the OAI protocol specification.  We have one recurring problem
> with XSV: it tries to open and parse xmlns attribute values.
> 
> It is my understanding that xmlns=foo is just a namespace.  It does not
> have to be a URL, it does not have to
> be locatable.

I'm not sure what you mean by "just" a name, but yes, it is a name,
just like http://www.w3.org/ is a name for the homepage of W3C.

It does not *have* to be locatable any more (or less) than the value
of an <a href="..."> attribute; i.e. it sure is nice when it
works, but the world is messy, and we have to be prepared for
404 errros.

404 errors occur about 6% of the time, according to some
web characterization stuff I read a while ago. That seems
to be just tolerable. But if it were the other way around...
if lookup failed 94% of the time, the web wouldn't be
very interesting.

Compare with the credit card system: as long as fraud
stays below about 2%, the system works. But if everybody
charged everything back all the time, the credit card
system would collapse.

>   A repository I am currently looking at has an info.pdf
> as its namespace!

If that's how you want to document your namespace, that's
your choice, but...

>  That is legal, but XSV
> tries to fetch it and parse it as XML.

... if you want to take advantage of the way XSV works,
you'll make an XML schema available there.
(you can make PDF available there too; use content
negotiation.)

> I do not believe you should be checking the xmlns attribute values.

I think it contributes to a self-describing web, which is a
more powerful web than one where looking up names fails
most of the time.

> Sincerely, Donna Bergmark
-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Friday, 8 February 2002 16:17:42 UTC