- From: Koen Holtman <koen@win.tue.nl>
- Date: Mon, 8 Sep 1997 21:23:34 +0200 (MET DST)
- To: Scott Lawrence <lawrence@agranat.com>
- Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
Scott Lawrence: > > In principle, I would have supported the less restrictive rule that > any header field can be sent in the trailer unless it is > specifically required to be in the header, but I think it is very > late in the game to be changing the rules on the client developers > now. I agree. I propose that we put a list of `headers which may be in the footer' in the 1.1 spec. We can further add a note to future protocol authors, telling them that whenever they define a header which is allowed in the footer of a HTTP/1.1 message, they should explicitely say so, because the HTTP/1.1 default is to disallow a header to be in the footer. Did we already discuss the problem of a proxy which gets a chunked 1.1 response and forwards it unchunked to a 1.0 recipient? It seems to me that, if we don't define something different explicitely, this proxy would be obliged to move all chunked footer-headers to the 1.0 message header before forewarding the response as a 1.0 response. This obligation would be a pain because it requires buffering of the whole response. So we should either allow a proxy to drop the footer-headers when converting to 1.0, or there should at least be some advance notification that these headers will be present, so that buffering can be avoided most of the time. The first alternative seems best to me. >Scott Lawrence EmWeb Embedded Server <lawrence@agranat.com> >Agranat Systems, Inc. Engineering http://www.agranat.com/ Koen.
Received on Monday, 8 September 1997 12:26:23 UTC