[Top] [All Lists]

Re: I-D ACTION:draft-ietf-sieve-vacation-00.txt

2005-03-21 12:01:05

I'm rather opposed to this extension as is.  I see a possibility for
enhancements here:
How about requiring vacation-supporting implementations having the
ability (where possible, i.e. everywhere but on the MUA) to *reject* a
message while sending a short message in the reject string or referring
the sender to a web-page?

Because this isn't what the extension is for. This sort of thing is something
best done with the notify extension.

I personally dislike vacation messages but they are undeniably quite popular.
IMO the community is best served by having a  mechanism that is specifically
tailored for this case. Attempting to handle all sorts of other cases is a
really bad idea IMO.

Also, why not change some SHOULDs to MUSTs, e.g. the ones in 3.6?

There are only two SHOULDs in that section. The first calls for responses not
to be sent to messages containing list- fields. AFAIK there's no specification
anywhere that clearly associates the semantics of list- fields specifically and
exclusively with list messages, so I think SHOULD is appropriate here. (In fact
it may be a bit too strong.) Calling out a specific table of list- fields known
to be the exclusive province of list-processed mail might resolve this and allow a MUST, but does not at the expense of making the specification less

The second SHOULD calls for messages not to be sent in response to messages
containing an auto-submitted: field with any value other than "no". This has a
similar problem: We have no idea what future  values will be defined for this
field, so again a SHOULD seems appropriate. And again the alternative would be
to specify the values for which a reply MUST NOT be sent, but again
at the expense of flexibility.

Look, we have little choice but to trust implementations to get this stuff
right. Either people pay attention to the specifications and their mandates,
and absent a compelling argument to the contrary a SHOULD is effectively the
same as a MUST, or they don't, in which case nothing we say will make the
slightest bit of difference.