- From: Roy Fielding <fielding@beach.w3.org>
- Date: Tue, 25 Jul 1995 19:06:47 -0400
- To: Larry Masinter <masinter@parc.xerox.com>
- Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
>I became a convert to the 'boundary' method for delimiting otherwise
>unbounded data, as is done with multipart/* messages. E.g.,
>
>content-transfer-encoding: bounded-binary; boundary="xxxxxx"
>
>would signal that the actual data started with
>
>--xxxxxx
>
>and ended with
>
>--xxxxxx--
>
>This is simple for both the sender and recipient; the packetized
>encodings seem messy. I'd worry that we'd need an
>'accept-transfer-encoding' to allow negotiation of CTE, though.
In fact, the "--xxxxxx--" does not necessarily end the multipart
body, it is just the end of the interesting stuff. Without additional
contraints on accuracy, multipart is worthless as a size delimiter.
In addition, it requires the receiver to scan the entire message
body for the delimiter, which has proven to be contrary to the design
goals of HTTP.
....Roy T. Fielding Department of ICS, University of California, Irvine USA
Visiting Scholar, MIT/LCS + World-Wide Web Consortium
(fielding@w3.org) (fielding@ics.uci.edu)
Received on Tuesday, 25 July 1995 16:09:47 UTC