- From: Larry Masinter <masinter@parc.xerox.com>
- Date: Tue, 12 Sep 1995 13:55:43 PDT
- To: FisherM@is3.indy.tce.com
- Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
> What we have here is a classic engineering tradeoff of 'quick negotiation > time' vs. 'high-user-display quality documents' vs. 'high-server-display > quality documents' vs. 'protocol extensibility'. If we allow for 100 MIME > types, where my 100 MIME types are different from your 100 MIME types which > are different from that server's 250 MIME types AND we want an extensible > protocol rather than a predefined list, it is going to take longer to do the > content negotiation. It is a fact of life. (Engineering Law: Good, Fast, > Cheap: Choose Any Two.) > Just thought of this off the top of my head -- would compressing, then > ASCII-encoding (like UU or base64) the Accept list help? Any feel for that? Did you miss the suggestion that clients hash all 'content-type-determining headers' and send them as a 'accept-hash:' instead? I suppose I'm choosing Good and Fast, at the expense of a little extra implementation complexity. One way to think of this is that hashing is a kind of compression mechanism -- you can compress any amount of data into 128 bits, but decompression can be very slow and take a large amount of communication.
Received on Tuesday, 12 September 1995 14:06:32 UTC