Re: Re: Opacity and mailto: in conflict

Tim Bray <tbray@textuality.com> schrieb am 22.09.2003, 16:27:22:
> 
> Norman Walsh wrote:
> 
> > In section 2.2, URI Opacity, we say:
> > 
> >   Although it is tempting to guess at the nature of a resource by
> >   inspection of a URI that identifies it, this is not licensed by
> >   specifications; this is called URI opacity.
> > 
> > Then later on we say
> > 
> >   mailto URIs identify mailboxes; ftp URIs identify ftp files and
> >   directories; etc.
> > 
> > It seems to me that these two statements are in conflict. Either you
> > aren't allowed to guess the nature of a resource from its URI, or you
> > are: it can't be both ways.
> 
> Not in the slightest.  It is perfectly OK for software to look at the 
> URI scheme and act on that basis, the semantics of URI schemes are 
> well-documented.  The problem is looking into the opaque part, i.e. 
> assuming that http://www.tbray.org/ongoing/Biz is a directory, or that 
> http://example.com/foo.html yields HTML when dereferenced.  Does the 
> spec need to be clearer on what's OK and what's not? -Tim

Tim,

as a pure consumer of the spec I'd say, yes, it is not clear.

Here is what I *think* about URI opacity and URI schemes:

- the semantics of the resource (what the nature of the resource is)
  may not be inferred from the URI
- the way to interact with the resource may be inferred from the
  scheme.

So, given a ftp:// URI software can invoke FTP operations on the
resource but may not infer that the resource's nature is 'FTP file'
(whatever that means).

So, if I am right, then

   "mailto URIs identify mailboxes; ftp URIs identify ftp files and
    directories; etc." 

is misleading because one may understand from this that there is a
semantic (as opposed to purely technical/software oriented) class of
resources that is called 'mailbox' and that all ftp:// URIs are
instances of 'mailbox'.

Or is my understanding of the issue wrong?

Jan
-- 
Jan Algermissen                <algermissen@acm.org>
Consultant & Programmer

http://www.topicmapping.com
http://www.gooseworks.org

Received on Monday, 22 September 2003 10:52:12 UTC