W3C home > Mailing lists > Public > uri@w3.org > June 2001

Re: fyi: should URIs convey protocol/service layering?

From: Roy T. Fielding <fielding@ebuilt.com>
Date: Fri, 15 Jun 2001 23:05:53 -0700
To: Dan Connolly <connolly@w3.org>
Cc: uri@w3.org
Message-ID: <20010615230553.C5945@waka.ebuilt.net>
On Fri, Jun 15, 2001 at 09:03:56PM -0500, Dan Connolly wrote:
> I'd prefer that they found out some other way than peeking
> in the name. But maybe there's no way to do that; there
> certainly wasn't at the time.

I can't think of any.  Even DNS tricks aren't really suitable because most
apps just use gethostname and have no access to the other DNS records.

> > In any case, http and https define two entirely different naming authorities,
> > even when their implementations reside on the same machine.
> 
> Really? Hmm... I'm not sure what you mean by that.
> Do you mean that the authority comes from the PKI
> certificate hierarchy, rather than the DNS hierarchy?

I meant that they default to different TCP ports, and the port is part of
the naming authority (witness the significant difference between "official"
web servers run on port 80, and non-official ones commonly run on port 8000).

However, I suppose we could claim that

    http://example.com/place/foo
    https://example.com:80/place/foo

refer to the same naming authority.  *shrug*  That kind of thing is considered
bad practice, since secure http services shouldn't share the same runtime
environment as non-secure services.

....Roy
Received on Saturday, 16 June 2001 02:08:59 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:25:03 UTC