This directory contains the implemetation for a new filtering action FAKESEND.

The purpose of that command is to let me send messages from home to work,
where they are resent locally. I do this because I usually forward my
private mail to home during nights and week-ends, but I don't actually
wish to make my home address public.

The way I use this is simple. I run Linux at home, and use MH to handle my
mail. I have a very special repl alias which configures the header to:

	Subject: Re: blabla
	References: <whatever>
	In-Reply-To: as appropriate
	To: Raphael_Manfredi@hp.com
	X-trigger: fakesend
                                    (blank line, end-of-real-header)
	To: real-to-header
	Cc: whomever
	X-Additional-Info: as appropriate (Organization, whatever)
                                    (blank line, end-of-additional-header)
	The body of my message start here.

Then, at work, I have this simple rule:

	X-Trigger: /^fakesend/		{ FAKESEND; DELETE };

And the FAKESEND command extract only some pre-arranged fields from the
real header (Subject, Date, References, and In-Reply-To), then gets
the additional header information, generates new From line from the
user information in ~/.mailagent if none was supplied in the additional
header, and passes the whole thing to sendmail -t.

WARNING: Don't use this X-Trigger rule as-is. Use something of your own
and don't tell anyone about it, otherwise other people will be able
to send you e-mail with this header and trigger FAKESEND, causing you
to appear to say something you would have not said otherwise, which
could be embarrassing. The mere existence of FAKESEND is yet another
proof that non PGP-signed e-mail cannot ever be taken as an evidence
in court that you said something.

NOTE: This version does not use the mailer configured in ~/.mailagent.
It hardwires the sendmail path, and currently relies on the -t option to
parse the new To and Cc header to know whom this message is to be sent
to. A possible improvement would be to do the parsing ourselves, which
would allow to honour the configured mailer from ~/.mailagent.

All this is automated. I don't actually generate the faked header by hand,
because I would make mistakes otherwise (like forgetting the X-Trigger line)
and that would be painful anyway.

Instead, I use MH... The local mh_profile file reproduces the relevant
settings for 'frepl' and 'freplq', aliases to 'repl' for "fake reply" and
"fake reply with quoting". The freplcomps file lies in my ~/mh directory.
When I want to send a faked quoted reply, all I do is 'freplq'. I then
proceed as usual. Isn't that magic?

NOTE: The freplcomps file is written in such a way as-to ignore the Reply-To:
field in the header. That's because some of the mailing lists I forward
are broken and set the Reply-To: to the list explicitely, which is bad.
Read http://garcon.unicom.com/FAQ/reply-to-harmful.html to know why.

The only thing that bothers me is that the local folder copy is made with
the "faked" form, not the real form the user will get (i.e. with the extra
header removed in place and inserted as part of the real header). I don't
know how to let MH do that. Any MH expert reading this and having a clue?

	Raphael Manfredi <Raphael_Manfredi@hp.com>
	Grenoble, France, March 30th 1998.
