post(8) not reading the profile was a long-standing deliberate design
decision, and the way the code is implemented it's not possible to
distinguish between "switches from the profile" and "switches from the
command line".
This is where it shows that you know the code base and I don't. :-)
Although I don't fully understand this; post already accepts -port on
the command line, the problem is (just) that I can't take advantage of
that because I'm not running post directly.
Putting things in mts.conf is also a pain; we never really figured out
a reasonable syntax for port number specification
I don't understand why the syntax would be difficult, though that's
probably only because I'm not familiar with the issues involved.
Still, from the perspective of an outsider who may be unaware of an
obvious reason why this would be a bad idea, what I'd propose is:
- the (new, optional) mts.conf entry would be specified as
port: NNNN
for some integer NNNN
- the entry would be ignored unless mts has a value for which a port
number is appropriate
- the specified port number would replace 587 as the default value for
post, to be overridden if -port NNN is supplied on the command line
Since rcvdist is deficient here, seems to me the right answer is to fix
that.
I agree. My point was just that there seems to be some disagreement about
how best to do that, and that it might be nice to be able to take the time
to discuss it thoroughly enough to reach a consensus.
But yes, fixing rcvdist properly (for an agreed-upon value of 'properly' :-)
in time for 1.7.1 would be preferable.
- Steven
--
___________________________________________________________________________
Steven Winikoff |
Concordia University | "Any teacher who _can_ be replaced by a
Montreal, QC, Canada | machine, _should_ be."
Steven.Winikoff@concordia.ca |
| - Arthur C. Clarke
--
Nmh-workers
https://lists.nongnu.org/mailman/listinfo/nmh-workers