Graham Klyne writes:
Archives that exist and work (as does the scheme described by Martin)
are infinitely more functional than ones that don't.
Isn't that the point of "running code..."?
Work is the key word, and the meaning of work depends on context.
Archives on the web must work in the context of the web, IMNSHO. And
similarly, archives pointed to by message/rfc822 header fields
must/should work in the context of message/rfc822 handling programs,
because those are the programs that look at such fields.
(That doesn't necessarily mean that the retrieved object must have a
certain content-type. I can imagine other ways to achieve the same