Re: [Extra] Email header / address parsing

[Arnt Gulbrandsen <[e-mail]>]

We've discussed this sort of this before, and I don't see what's different 
this time. If the server can parse it well enough to present the correct 
domain with ".invalid" appended, why not present the domain without 
.invalid? If the server cannot parse it that well, why even try to include 
that address in the list?

IMO, the best way to handle this is to say: "If a header contains syntax 
errors, whether major or minor, the server MAY present a best-effort 
ENVELOPE or BODY. This response may be be either incomplete (the server 
drops the addresses that could not be parsed and presents as many as it can 
of the others) or unreliable (if an address is invalid, there may not be a 
single obvious way to render it valid)."

This is not a compatibility break, doesn't add new requirements, does 
suggest a way to handle some syntax errors, and it doesn't expose clients 
to new expectations: Four out of four, say I.

Arnt