- From: <touch@isi.edu>
- Date: Tue, 18 Feb 1997 17:10:08 -0800
- To: masinter@parc.xerox.com, ses@tipper.oit.unc.edu
- Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
> From http-wg-request@cuckoo.hpl.hp.com Tue Feb 18 16:54:32 1997 > Resent-Date: Wed, 19 Feb 1997 00:49:58 GMT > Date: Tue, 18 Feb 1997 19:49:27 -0500 (EST) > From: Simon Spero <ses@tipper.oit.unc.edu> > On Tue, 18 Feb 1997, Larry Masinter wrote: > > > I think the short answer to "how to add new protocols to HTTP" > > is that "there isn't any way right now" and that if anyone > > has a serious proposal for how to do so and under what circumstances > > it would be a reasonable thing to do, they should write it up. > > Techniques available: > > 1) try alternate protocol, and use it if it works. Cache result > of test. Increases the connection latency. Bad in general. Besides, which protocol do you try? > 2) Use DNS records to get info on which protocol to use. I can't do the DNS lookup until I can parse the //.../ location field, which I can't do until I know the protocol. Circularity ensues. > 3) Use Upgrade: header Two connections are better than one? > 4) Use invalid method to probe. Same? The HTTP spec is just underpowered on this one. The current URL starting with http: assumes TCP, and makes lots of other assumptions about the interpretation of the location id. Joe ---------------------------------------------------------------------- Joe Touch - touch@isi.edu http://www.isi.edu/~touch/ ISI / Project Leader, ATOMIC-2, LSAM http://www.isi.edu/atomic2/ USC / Research Assistant Prof. http://www.isi.edu/lsam/
Received on Tuesday, 18 February 1997 17:15:24 UTC