- From: John Franks <john@math.nwu.edu>
- Date: Mon, 27 Jul 1998 09:21:38 -0500 (CDT)
- To: Dave Kristol <dmk@bell-labs.com>
- Cc: HTTP Working Group <http-wg@cuckoo.hpl.hp.com>, jg@w3.org, http-wg@hplb.hpl.hp.com
On Mon, 27 Jul 1998, Dave Kristol wrote: > Last week I wrote: > > > > I propose here corrective wording for open issue RANGEDELIM (described > > in <http://www.ics.uci.edu/pub/ietf/http/hypermail/1998q2/0141.html>). > > [...] > > Now, HTTP introduces an interesting ambiguity, because the first line > > of an entity body could well be considered the beginning of a line, and > > it follows a CRLF. However, since we must ignore the CRLF that > > separates the headers from the body, the body does not really begin > > with a CRLF. So by my reading of RFC 2046, an HTTP multipart entity > > would have to include an extra CRLF preceding the boundary. I have > > amended Section 19.2 of the HTTP spec. accordingly, along with the > > example. > > After discussions over the weekend, especially exchanges with Ned Freed, > co-author of RFC 2046, I am amending my proposed changes. In > particular, most of them go away. It turns out that a careful reading > of RFC 2046 shows that an extra leading CRLF is unnecessary, and that > the HTTP example was correct. (Details: the multipart-body grammar > begins "[preamble CRLF]", which makes the cruft that often precedes > multipart bodies, plus the CRLF that ends it, optional.) > Enough people have been confused by this that it is important to have an explicit warning that the leading CRLF is allowed and optional. John Franks john@math.nwu.edu
Received on Monday, 27 July 1998 12:13:06 UTC