httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Roy T. Fielding" <>
Subject Re: cvs commit: apache-1.3/src/support httpd.exp
Date Sat, 02 Jan 1999 03:05:47 GMT
>mod_negotiation isn't the only module that can do negotiations... what
>happens if a sub request also has a vlist_validator because it was
>negotiated?  (Maybe I don't understand vlist_validator... maybe
>you could put a couple more notes into httpd.h ;)

About all I can tell you is that a vlist_validator is defined in RFC 2295
as a way of providing an etag for the negotiation mapping defined by
the Alternates header field, such that caching proxies can use it to
perform server-side negotiation on behalf of the server.  The effect on
Apache is that it forces the responses to become invalid whenever the
mapping is changed.  A rational design would have put this information
in the Alternates field value and not in Etag.

In other words, it is an excessively controverted feature that will never
be used because a caching proxy is the least capable application in the
request chain for making correct content selection decisions.  TCN has
no clue about negotiation mapping hierarchies, so it wouldn't surprise
me at all if it is broken for negotiating sub-requests.


View raw message