Date: Tue, 23 Mar 1999 16:09:57 -0500
From: Gregory Sereda <gsereda(_at_)maillennium(_dot_)att(_dot_)com>
My preference would be stronger than SHOULD NOT but not quite
MUST NOT. I guess that means stating the discard SHOULD NOT
generate a NDN. Perhaps, with text stressing that the usefulness
of discard is greatly disminished by implementations that generate
a NDNs. The developer should try very hard to implement discard
correctly.
Perhaps we need to consider adding REALLY to the KEYWORDS document
(REALLY REALLY SHOULD NOT).
Wonder if that can be written up for next Thursday...
I think there is still value in a sieve implementation that
cannot implement discard without a NDN, but users of the
implementation would just have to live with its limitations.
I would rather have discard be optional than this. If discard is
equivalent to reject, it is not discard anymore. Having two meanings
for discard is as bad as two meanings for fileinto.
--
Tim Showalter <tjs+(_at_)andrew(_dot_)cmu(_dot_)edu>