--On 2004-1-23 2:33 PM +0000 Paul Smith <paul(_at_)pscs(_dot_)co(_dot_)uk>
wrote:
What I'd quite like is a way of registering a header field (including
basic semantics, syntax and an ABNF definition (for many existing X-
fields this could be trivial with some examples to work from)). This
would go to a group of 'learned volunteers' who make sure it's not
abusive (ie not 'grabbing' a useful header field name for an
inappropriate purpose, or not commercially advantagous to someone to
the detriment of others etc) and that's it, registered. The 'learned
volunteers' may also give feedback to the registrant saying something
like 'we think this is a really good idea, but needs developing in
such and such a way', and then they reserve the header field name, to
give the full registration time to be improved.
"Header hijacking" is also one of my major concerns, and I'm in favour
of what Paul is suggesting.
If X-* dies I suppose the experimental use folks could use something
like a base64-encoded hash of the proposed header name (or any other
method that would generate random strings with a suitably low
probability of namespace collisions).
--lyndon