On Thu, 2005-04-07 at 08:41 -0700, ned(_dot_)freed(_at_)mrochek(_dot_)com wrote:
I suggest
Vacation responses are not just per address, but are per address and
reason argument.
I think that makes it clearer that the intent is to consider the
verbatim unexpanded reason argument.
I agree, however, don't we also want to require that the calculation to
include
the :subject argument?
okay, so here are all the possible arguments:
Syntax: vacation [":days" number] [":subject" string] [":from"
string]
[":addresses" string-list] [":mime"] [":handle" string]
<reason: string>
we could the split depending on whether the option has an influence on
the sent message. that means only :days and :addresses are
ignored. :mime is unclear, but you shouldn't have two invocations with
the same reason, one with :mime and the other without. (actually, you
_can_ do that if the MIME bits are put in a variable, so let's include
the existence of :mime in the hash, too.)
that said, I'm not particularily concerned about the specification
causing too few messages to be sent. after all, the case we're
discussing is how many reminder messages should be sent. the sender
will always receive at least one vacation message.
I'm leaning towards keeping it simple, particularily since the user
explicitly can get the behaviour he wants using :handle. in other
words, stick to just the reason.
I believe there was a time when we didn't want to refer to variables in the
vacation draft bevcause it was assumed variables would take much longer to
complete. That's no longer true - variables is in last call while we're still
fine tuning vacation. So how about simply referring to the variables document
and say that any hash SHOULD cover the unexpanded :subject and reason
arguments?
sounds good to me.
--
Kjetil T.