On Jan 04, 2015, at 03:10 PM, Jeff Breidenbach wrote:

So the homebrew hash was dumb idea, and will gracefully retire in favor of the standard mailman hash algorithm. It was pretty neat to see some usage, though. And Archived-At still looks like a winner.

Definitely. X-Message-ID-Hash was just a way to elaborate on RFC 5064 $3.2, and because Base 32 was chosen, there isn't any need to worry about needing to escaping non-ascii characters in the Message-ID, nor case/letter/number ambiguity, and still have a reasonably short-ish, human readable URL. Mailman guarantees that any message it forwards will have a Message-ID, although it's usually the case that something upstream will also make such a guarantee.

The X-Message-ID-Hash is a pure convenience. We could omit it since we still create and add an Archived-At header with the hash as the last path component, and of course with a well-documented algorithm, it can always be calculated from the Message-ID. It's just handy to include it for other system components that want to combine it easily with RFC 2369's List-Archive header.

Cheers, -Barry