RE: comments on RFC2518bis-02, sec 6.3

> From: Lisa Dusseault [mailto:lisa@xythos.com]
> Sent: Sunday, June 29, 2003 7:50 PM
> To: 'Julian Reschke'; 'Jason Crawford'
> Cc: w3c-dist-auth@w3.org
> Subject: RE: comments on RFC2518bis-02, sec 6.3
>
> > Yes. A non-registered URI scheme doesn't have *any*
> > guaranteed uniqueness, so it doesn't serve it's stated purpose...
> >
>
> That's not true.  If I create a URI scheme where the scheme name is
> "http://www.xythos.com/storageServer/locktoken/", without registering this
> with the IETF, it can still meet the uniqueness guarantee.

You can't create a URI scheme with that kind of scheme name. The scheme part
of a URI is the string before the first colon, so the URI scheme for that
URI is just "http" (which indeed *is* a registered URI scheme).

And yes, you can use these kinds of URIs for lock tokens.

> For that matter, a sufficiently long randomly generated set of characters,
> as long as it meets the URI formatting requirements,
> statistically meets the
> uniqueness guarantee.

How do you "statistically" meet a requirement?

Julian

--
<green/>bytes GmbH -- http://www.greenbytes.de -- tel:+492512807760