Re: [Nmh-workers] Conflict between "mime" command and attach

[Ken Hornstein]

>To deal with the false directives when using attach:
>WhatNow could look for any lines that begin with # during
>the first call to "attach", and warn or refuse.  Those who
>use # for replied-to text would have to find something else.
>Or we could get more sophisticated, back to the "challenging"
>aspect above.

Okay, this sort of puts my finger on the issue in my mind.

The reason this is cropping up now is that we want to get to the
point where a MIME headers are always generated (I assume this is
non-controversial).  Using the tools we have, this means running
mhbuild.  This was never designed to be run all of the time, hence the
issues with directives.

The reason all of these various suggestions regarding putting mhbuild
directives in the text feel wrong to me is that it BY DEFAULT assigns
special meaning to the message body where there was none before.
That seems like a major change, and just feels wrong.  I know, David
suggested looking for false mhbuild directives, but in addition to the
technical challenges that just feels wrong to me; trying to guess what
is and isn't a directive suggests to me that we're doing the wrong
thing.  It's fine for users who WANT to create mhbuild directives, but
it just Seems Wrong that message bodies now assign special meaning
to '#' at the beginning of the line.  That would mean we're now
unilaterally switching from users creating a "message draft" to a
"mhbuild composition file".

In contrast, the current implementation of attach that creates a special
header actually feels "more" right; creating headers and having the
tools use them to process a message is more the MH way, as we've seen
time and time again.

Can the people who want to have "attach" append mhbuild directives
explain what their thinking is, specifically why they think their
approach is preferrable?  I went back and looked at the thread very
carefully, and none of the proponents of this approach really covered
why they thought this was better.  Ok, Ralph did say that he wanted to
look at the headers post-MIMEification to adjust them; I kinda feel
that's an "expert mode" feature and people who want to do that are
probably comfortable with mhbuild directives.  I really do want to
understand people's arguments before I make a decision.

(It occurs to me that the logic to do the MIME to filetype mapping that
is done in post really should be moved into mhbuild, as that's more
mhbuild's job.  Just an implementation detail, though).

--Ken