Re: Getting full URI to the server

At 5:55 AM 2/13/95, Roy T. Fielding wrote:
>Well, after reading all the comments, I have come to the following
>   1) Too many web developers work on the weekends;

Agree. (unfortunately)

>   2) Marketing has too much influence on web technology; and,


>   3) There's not much I can do about either one.

Disagree! You're working very hard to put together a good standard. If it
meets the needs of developers and users and gets accepted in a timely
fashion, this will have much more influence on the WWW than rogue
commercial development organizations.

>So, which one should we include in the 1.1 specification:
>   Host:

As you mentioned earlier, this is all that is *required*, since everything
else should be in the URL as received. It is certainly is easier to
implement because no parsing is required to extract this info.

>   Orig-URI: http//

On the other hand, this format is a little more work to parse, but contains
valuable information that is often lost or munged when data passes through
proxies, etc. Looking at just long-term usefulness, this version seems to
have a greater chance at being able to support multiple, unforeseen uses.
(it certainly makes implementing proxy servers easier...)

>Also, should it be:
>   a) recommended for all requests


>   b) recommended only for requests to standard URLs like / and /site.idx


>I am assuming that it will not be recommended for requests that already
>use the full URI, and that will remain the goal for 2.0.

Why not be consistent? Proxies can still munge a full URI as easily as they
can mangle a current version. Might as well maintain it unscathed in the
header field where the real server can grab it.

Chuck Shotton                                                   "I am NOT here."

Received on Monday, 13 February 1995 07:40:40 UTC