- From: Josh Cohen (Exchange) <joshco@exchange.microsoft.com>
- Date: Sat, 4 Sep 1999 17:39:25 -0700
- To: http-wg@hplb.hpl.hp.com
In going over some implementation stuff here, I ran into an issue that I apparently misread in the spec. According to the spec, the host: header is always required in HTTP/1.1. In my mind, I always perceived that to mean that it was required as long as a relativeURI was used in the request. Looking closely at the spec, I see that is not the case. It seems to me that if you have an absolute URI, the host header is redundant. If I remember correctly, from the wg meetings in the long past, the host header was added because absolute URIs werent used to origin servers. Now that HTTP/1.1 allows the use of absolute URIs, doesnt it seem wise to remove the requirement of the host: header when they are used? It certainly makes things simpler if Im build a new client/server system to just use absolute URIs consistently. Can anyone explain what the justification was for requiring the host: header even when their is a perfectly good host indication in the absoluteURI? Frankly, I can't beleive I didn't raise this sooner, I always assumed this was the case. I only ran into it in a debate here with a colleague.
Received on Saturday, 4 September 1999 17:43:00 UTC