Re: Mutt dependency on an MTA
On Sat, 31 Jul 1999, Hamish Moffatt wrote:
> I don't feel strongly enough about this topic to reply to each of your
> points. I agree that some MUAs (specifically pine and netscape) are
> operable without a local MTA. I do not see that this warrants any
> action ie modification of packages. For most users, an MTA is required.
This actually is the definition of Recommends:, not Depends:. A Depends:
is something that IN ALL CASES the lack of will cripple the dependent
package, something that we both agree won't happen across the board for
all MUAs.
>
> I would consider a UNIX system incomplete without one. At work
> we have UNIX systems as embedded controllers and even those run sendmail.
Okay, make it Required--making it a Depends: on a possibly Extra package
in that case is ludicrous. BTW you're right--a *nix without a MTA is
crippled, I just don't see a MUA without a local MTA as crippled.
> I don't see how having an MTA installed unnecessarily adds an
> additional point of failure; pine and netscape store their mail in
> private directories, so damage to /var/spool/mail would be inconsequential.
I didn't say it was an additional point of failure, I said that it was a
kludge: interfacing things that really didn't need to be interfaced. It
adds an additional point of inelegance, no more, no less. I just happen
to prefer elegant solutions: maybe we can have Galt's razor: "Where there
are two equally workable solutions to a problem, the more elegant one
should be used" :) Now as for the "equally workable" part--you and I
disagree whether the two solutions are equally workable: so? That's why
it's a dialouge: two monologues would get pretty boring. :)
>
> Hamish
> --
> Hamish Moffatt VK3SB (ex-VK3TYD).
> CCs of replies from mailing lists are welcome.
>
Artificial intelligence is no match for natural stupidity.
Who is John Galt? galt@inconnu.isu.edu, that's who!
Reply to: