W3C home > Mailing lists > Public > www-html@w3.org > June 2000

Re: https:/ and http:/ using relative URLs to switch between secure and non-secure

From: Roy T. Fielding <fielding@kiwi.ICS.UCI.EDU>
Date: Tue, 20 Jun 2000 17:00:56 -0400 (EDT)
To: Dan Connolly <connolly@w3.org>
cc: EF <energyflow@energyflow.com>, www-html@w3.org, uri@w3.org
Message-ID: <200006201400.aa02602@gremlin-relay.ics.uci.edu>
>(b) I believe this request was considered and declined during the
>development of the URI spec, RFC2396, though I don't have
>specific pointers handy. I think libwww used to implement
>it... the part about
>	scheme1://auth1/path1 + scheme2:/path2 = scheme2://auth1/path2

libwww only ignored the scheme if both schemes were identical.

>This is the first time (I think) that it's been proposed to allow:
>
>	scheme1://auth1/path11/path12 + scheme2:../path2
>		= scheme2://auth1/path11/path2
>
>In any case, I doubt you're
>going to get a change to the relative URI parsing
>algorithm at this stage. It's just too widely deployed.

It also wouldn't make any sense to do that given that not all URI
use the //hostname syntax and therefore such a notion would cause
the relative URI parsing to be scheme-dependent.  That would suck.

....Roy
Received on Tuesday, 20 June 2000 17:24:37 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 27 March 2012 18:15:43 GMT