W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > September to December 1997

Re: HTTP/1.1 ISSUE: TRAILER_FIELDS - Proposed Resolution

From: <rlgray@raleigh.ibm.com>
Date: Wed, 19 Nov 1997 17:48:33 EST
Message-Id: <199711192248.RAA24380@rtpmail02.raleigh.ibm.com>
To: HTTP Working Group <http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com>
X-Mailing-List: <http-wg@cuckoo.hpl.hp.com> archive/latest/4749
Being a server implementer, not a client implementer, I fail to see how
the Trailer proposal solves the case-by-case problem.
It does, in fact, list (currently 3) fields that MUST NOT occur in the
trailer.  This seems to me to be unnecessary.
Perhaps you could provide an example of what could break by a client
expecting a header field in the header, not finding it there and then
finding it later in the trailer without a forward declaration?

And while I am at it, if you are going to abbreviate
"Accept-Transfer-Encoding" as "Accept-TE", why not shorten it to
"A-TE"?  I seem to remember this comes up from time-to time, but the
fact is, HTTP _is_ a human-readable protocol.  I see only two headers
in 2068 that contain abbreviations, Content-MD5 and WWW-Authenticate.

I tend to think that since there is some symmetry between the "accept-"
header fields and the "content-" header fields, and that all of the
Accept- header fields refer to the entity desired, not the format of
the message, another header name should be used; perhaps using the
Expect header field with "transfer-encoding-deflate" or just "deflate"
would be acceptable (err, pun intended ;-)

Or, am I just particularly dense today?

** Reply to note from Henrik Frystyk Nielsen <frystyk@w3.org> Wed, 19 Nov 1997 14:48:27 -0500
> At 14:02 11/19/97 EST, rlgray@raleigh.ibm.com wrote:
> >I don't see the point of the trailer field at all.
> >Why not just say, as suggested:
> The purpose of the Trailer header field is to provide a general mechanism
> describing where headers are in a message, so that it doesn't have to be on
> a case by case basis (see also description of problem [1]).
> The notion "Only put fileds in the trailer that can be safely ignored by the
> client (e.g., Last-Modified)" doesn't really say much as it depends on who
> you are (as a client) and what you are doing.
> The combination of Accept-TE and Trailer allows the server to put headers
> in the trailer and the client to say no thanks.
> Henrik
> http://www.ics.uci.edu/pub/ietf/http/hypermail/1997q3/0659.html
> --
> Henrik Frystyk Nielsen,
> World Wide Web Consortium
> http://www.w3.org/People/Frystyk

Richard L. Gray
will code for chocolate

cc: Henrik Frystyk Nielsen <frystyk@w3.org>
Received on Wednesday, 19 November 1997 14:50:12 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:40:21 UTC