- From: Koen Holtman <Koen.Holtman@cern.ch>
- Date: Tue, 16 Mar 1999 15:38:24 +0100 (MET)
- To: Larry Masinter <masinter@parc.xerox.com>
- cc: Keith Moore <moore@cs.utk.edu>, web@apps.ietf.org, discuss@apps.ietf.org
I just read the draft. I think the basic idea of putting a small HTTP message as the payload in a broadcast UDP message is rather cute. But it will only remain cute for me if the HTTP message size stays within the maximum safe/generally supported size of a single broadcast UDP message. I don't know what this size is nowadays, I am getting conflicting data from various sources. Complex to-be-defined mechanisms for supporting arbitrary-length HTTP-over-UDP broadcast would spoil the idea for me. Anyway, I would not consider this to be something on top of HTTP, it is rather something that re-uses some of the extensible syntax in HTTP, which is entirely legitimate. As the use case is LAN based (or at least intranet based), I don't think the IETF should devote major resources to this. Koen. On Tue, 9 Mar 1999, Larry Masinter wrote: > > one of the other ADs ran across this draft, which layers a protocol > > on top of HTTP. > > It layers a protocol on top of something which is not > defined (yet, although it promises to): HTTP layered > over (multicast and unicast) UDP. > > I think HTTP layered over multicast & unicast UDP will > be exceedingly difficult to define correctly. HTTP messages > and responses are arbitrary length, and require one-for-one > responses to requests. > > >... we're looking for a few good HTTP experts to > > sanity check this draft > > The 'sanity' of this approach escapes me. > > Larry > > >
Received on Tuesday, 16 March 1999 09:39:44 UTC