Keith Moore <moore(_at_)cs(_dot_)utk(_dot_)edu> writes:
and note that I didn't say 'indefinitely'. but lots of people keep old
mail around. while it's hard to argue that people should keep buying
5.25" disk drives for the sake of reading their old floppies (partially
because the medium is essentially useless for storage of new things),
it's a bit easier to argue that MUAs should contain a bit of extra code
to handle slight differences in message format that were once widely
used.
That doesn't, however, imply that a standard should specify that behavior.
Consider this analogy: a few years ago, a new version of the C standard
was released. It has some incompatibilities with the previous version of
the C standard, although not a whole lot. But it is possible for
previously conforming programs to not compile under the new standard in
certain edge cases.
Does the new C standard mandate that all compilers be able to handle both
the new and old standard? No, of course not.
Do most compilers implementing C99 have multiple modes so that you can
tell them to process a source file as C89 or C99? Yes, in my experience.
Vendors understand the need for backward compatibility with their existing
install base all by themselves. There's no need to mandate that in a
standard. (Now, mandating the ability to understand *new* e-mail messages
that are still seen on the network, generated by old software, is another
matter and a proper focus of an IETF standard IMO. But mandating the
ability to understand old mail archives is not, again IMO.)
--
Russ Allbery (rra(_at_)stanford(_dot_)edu)
<http://www.eyrie.org/~eagle/>