Re: Getting full URI to the server

>From Roy T. Fielding:
> The final question is: Does the additional functionality justify the cost
> and effort of including the Host header in the 1.1 standard, with the
> necessarily strong recommendation that it be included with all requests?
> In my opinion, the answer to this last question is NO.

>From John Franks:
>I suspect that the ability to customize the default page based on
>hostname part of the URL is the single most requested feature from
>server maintainers.  I doubt that a week goes by without a thread on
>this subject in c.i.w.providers.  The practice of using multiple IP
>addresses on a single host for the sole purpose of working around this
>deficiency in the protocol is becoming increasingly common.

Though I can't find it right now, there is actually a web page/tutorial
devoted to using multiple IP addresses on one network interface for this

>Those who are critical of adding a new HTTP header just for "vanity
>addresses" should keep in mind that the likely alternative is the
>wasteful use of IP addresses just for vanity addresses.

Not only is it likely, it is happening quite a bit.  There are plenty of
businesses buying machines and connections exclusively to put up web
servers, and they often want to split costs with other businesses.

I agree with Roy that some protocol enhancement should drive the addition
of a Host: header.  It is currently possible to put more than one server at
one address by assigning one of them a non-standard port number
( and  A Host: header could
provide the same functionality in a transparent manner.  As Chuck Shotton
mentioned, this addition would require no change to servers.  The benefit
-- allowing providers to seperate offerings into categories without making
the user choose from a home page menu -- seems worthwhile.  Not just to

M. Hedlund <>

Received on Sunday, 12 February 1995 12:52:33 UTC