Re: Content negotiation

Shel Kaphan (sjk@amazon.com)
Thu, 9 Nov 1995 09:28:05 -0800


It does seem like getting something like what we've been talking about
to happen would best be done completely outside HTTP. The point has
been raised that if browsers had to communicate their own properties,
either in original requests or subsequent to redirects,
that would not account for properties added after the browser was
distributed (unless the browsers themselves looked up their properties
somewhere dynamically), and would also not account for bugs discovered later.

If we could get to the point of having some reliable, mirrored sites
that would serve a standardized database of these properties, that
would do it. I also like the idea of third parties being responsible
for keeping track of what browsers can and can't do. The databases
should probably be keyed by the User-Agent string, and return a
machine-readable document (in a format to be discussed later). This
requires no protocol extensions or cooperation of browser vendors, and
so is not hopelessly impractical.