Re: [Nmh-workers] rcvdist with non-default port

[David Levine]

Ken wrote:

> I ... do not think that is safe to put in?  It seems to me like that
> would fail if your command line looks like:
>
> rcvdist -server foo.bar -tls address@hidden address@hidden

I'm not sure what you mean.  How is this not safe?

    $ rcvdist -server foo.bar -tls address@hidden address@hidden </dev/null
    post: only one message at a time!
    /home/levine/lib/nmh/libexec/nmh/post: exited 1

> address@hidden would be swallowed and passed to post.  If you just
> had one address rcvdist would fail.

Right:

    $ rcvdist -server foo.bar -tls address@hidden </dev/null
    rcvdist: usage: rcvdist [switches] [switches for postproc] address ...

> You could maybe make sure that you
> always put a switch that takes an argument last, but that just seems
> lousy as well.

I don't see a need to do that.

It's up to the user to write an rcvdist command that does what
they want.  The same is true for other nmh commands that forward
arguments and switches to other programs, e.g., show -form 1 2 3.
(In that case, show(1) passes "-form 1" to its showproc.)

The way things are now (master and 1.7 branch), users can no
longer write an rcvdist command that does what they want.  One
fix, to pass along switch arguments to post as documented, is
straightforward.  And results in switch handling that is no
worse than for other nmh commands.

I don't think that we should attempt more invasive changes to rcvdist for
1.7.1.

David