Python mailing list vs. References: (was Re: My quarterly question on Design by Contract in Python...)

[Tim Peters]

[Barry A. Warsaw]
> ...
> But I agree with the sentiment that says Mailman should be mucking
> with headers as little as possible.  I don't think rewriting
> References: and In-Reply-to: falls under the category of "as little as
> possible".

I believe Aahz is suggesting:

1. Never change a References header if there is one.
2, Never change a In-Reply-To header if there is one.
3. Never create a In-Reply-To header if there isn't one.
4. If there isn't a References header, but there is a In-Reply-To that
   appears to hold a sensible msg ID, then-- and only then --synthesize
   a References header from that.

That's not as little as possible, but it does seem to be very little all the
same  <wink>.

purity-vs-usefulness-ly y'rs  - tim