- From: Paul Leach <paulle@microsoft.com>
- Date: Wed, 7 Aug 1996 15:25:25 -0700
- To: 'Jeffrey Mogul' <mogul@pa.dec.com>, 'Daniel DuBois' <dan@spyglass.com>
- Cc: 'http-wg' <http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com>
>---------- >From: Daniel DuBois[SMTP:dan@spyglass.com] >Sent: Wednesday, August 07, 1996 3:11 PM >To: Paul Leach; 'Jeffrey Mogul' >Cc: 'http-wg' >Subject: RE: Sticky headers and pipelining (was: Sticky header draft --as an >attachment) > >At 02:35 PM 8/7/96 -0700, Paul Leach wrote: >>The mechanism for that in the draft is that the client sends the header >>name and an empty value, which means that, even though there was a value >>for that header in the last message, this message should be interpreted >>as if the header with that field-name is not present in the message. >> >>There must be some mechanism like this, independent of the semantics of >>any request header, in order for sticky headers to work at all. >>[...] >>One could invent a header with the property that (say) >> Foo:<CRLF> >>meant one thing, and >> Foo: Yes >>meant another. I don't believe that are currently any examples of this, >>however. Certainly, either we would either have to outlaw adding any >>such headers to HTTP, or outlaw their participation in the sticky header >>mechanism. > >Uh.. The semantics of an empty Accept-Encoding are mentioned in the 11-06 >draft. And I think the semantics of "Accept:<CRLF>" are both self-evident, >and clearly not the same as a lack of the header. Certainly the BNF allows >numerous headers to be NULL. Oh, foo. Drat. Darn. Shucks. I'll have to look at them. Mumble. > >I was presuming (not having read the draft, can I admit that in a discussion >on the draft?) that your intentions were that a sticking client would >indicate turning off individual headers on subsequent requests by sending >the explicit format that was semantically equivalent to the default of those >individual headers. One can always do this. But is there a default for all headers? And is it concise enough to be acceptable? > As in, sending "Accept-Encoding:<CRLF>" since empty and >non-existant have the same meaning, or "Accept: */*" since that has the same >meaning as non-existant. This would mean that all stick-able headers must >have a explict default-equivalent format (User-Agent:<CRLF> ??), or a that >you need a global way to wipe off all the stickyness (Here's where I propose >the Towel: header.), or, finally, that you have a >"Stuck-Headers-That-Slipped-Off:" header. I was trying to avoid either of the last two approaches. But maybe I can't. Paul
Received on Wednesday, 7 August 1996 15:31:11 UTC