[NNTP] STREAMING diffs (take 2)

[Ade Lovett]

Russ Allbery wrote:
> This would also work for me.  (400 could also be listed as a response for
> dealing with excessive CHECKs.)

I still absolutely do not see the need to explicitly mention this,
unless EVERY possible case for doing bad things to a server is enumerated.

For example, with a number of pre-commit history file caches, STAT can
be trivially used to stomp all over the cache, rendering
lookup-via-message-ID slow elsewhere.

Likewise, depending on how an article count is generated, flip-flopping
GROUP requests could similarly bring a server to its knees.

The point I'm making is not so much an issue with the text itself, but
the case that if we do it here, then we would need to seriously extend
the document to cover other cases where the protocol could be abused.
We also then leave ourselves open to then having to deal with those
situations that aren't explicitly mentioned.

If someone(tm) wants to go ahead and write up a formal BCP on how
servers might want to interact in the real world when faced with such
protocol abuses, that would probably be a Good Thing.  But it does not
belong in the protocol description document itself.

-aDe