| Newsgroups |
gmane.ietf.usenet.format |
| Message-ID |
<[email protected]> |
"Forrest J. Cavalier III" <mibsoft@xxxxxxxxxxxxxxx>:
>That is because the server's record of message IDs is a
>finite list in practice.
>No one should offer an opinion on how to resolve #1416
>unless and until they understand this point.
Ignore that other draft behind the curtain, _I_ am the great and powerful
USEPRO! Believe what I say, even if we said something different before.
If this is point is so critical, then USEFOR should have said it, instead
of what it does say.
If you want to reopen USEFOR and debate whether "message id" or "message
id plus date" is the globally, chonologically unique identification for a
message, I'm game. If you don't want USEFOR to say "forever", then let's
debate it and make the change. We should not say "forever" if we don't
mean it, just like we shouldn't say "MUST" unless there is a valid reason.
No, there is no reason to add "date" to the message id. You keep a list of
message ids and the starting date of the list. You make the statement in
USEPRO that servers can ASSUME that any article with a Date prior to the
starting date of the history has been seen and is thus a duplicate. End of
problem. You assume the message id is unique, just add a different test
for acceptance.