I like the improved clarity in a lot of things, such as in the ordered
list of techniques that are tried for carrying out the reject action.
However I think the problem still remains that the script writer does
not get any control over which are tried. I mention again the potential
irony that whereas a script writer will intentionally avoid 'reject' if
they know it will result in a separate bounce message, the writer will
be tempted more often to use 'reject' if they think it may result in an
SMTP time refusal. The irony being that 'reject' will be then be used
more often, which without some influence as to how it falls back,
can result in more bounce messages.
Personally I liked the syntax proposed by Kjetil Torgrim Homme,
probably here:
http://www.imc.org/ietf-mta-filters/mail-archive/msg02408.html
i.e.:
reject :refuse :stop "Don't send e-mail to this address";
fileinto "INBOX.old-address";
stop;
where ":refuse" says "only execute this if you can do a protocol refusal"
and ":stop" says "stop if the action was taken."
If the script continues, the script writer can then add steps to further
classify the message, to see if a non-protocol refusal is still
warranted, or otherwise to file or discard the message.
I don't mean to belabor this, but I thought this had some support.
mm