Kjetil Torgrim Homme wrote:
On Thu, 2006-07-27 at 15:01 +0100, Alexey Melnikov wrote:
Kjetil, there is very much 50/50 split between people who want to change
the SHOULD to MUST and people who want to keep it as SHOULD or even
relax it.
RFC 3028 used SHOULD. Considering even split (and good technical
arguments on both sides) I don't see any reason to change it to anything
else.
yes, it says that in general:
Implementations SHOULD prohibit reject when used with other actions.
but it also calls the explicit combination we're talking about
"invalid":
Implementations might
might is non normative here. I certainly wouldn't interpret it as "MUST".
even go so far as to ensure that scripts can
never execute an invalid set of actions (e.g., reject + fileinto)
before execution, although this could involve solving the Halting
Problem.
I don't think changing it to a MUST is changing the spec, it's just
clarifying it.
I think part of the reason why we have this discussion now is because
the text you quoted is sufficiently ambiguous.