Re: [Nmh-workers] formatting headers - a weird case

[Peter Maydell]

Jon Steinhart wrote:
>Note that there are two Return-Path header fields.  If you do a scan -format 
>"%{Return-Path}"
>you'll get the value of the last one.  Is that what we want?

I don't know, but the BUGS section of the scan(1) manpage claims you get the
first one, not the last, so if that's wrong we should change it :-)

>What is the proper formatting of header fields that occur more than once.

I guess we could extend the mh-format(5) spec so that everything that
accepted a 'component' (header name) allowed 'component' (gives first or
last or anyway something consistent), 'component:all' (gives the whole lot
concatenated), 'component:n' for integer n (gives the nth instance) and
'component:-n' (gives the nth-from-last instance).

But that's quite a bit of work and I'd rather see a genuine use-case
for looking at multiple headers first :-)

> BTW, I haven't checked, but I'd guess that this is a memory leak!

We should fix that, at least, if so.

PS: one of those Return-Path: headers is in breach of RFC5322 because
it doesn't have the angle-brackets. Furthermore it is the fake one
cooked up by nmh as a way of showing the sender address and timestamp
from the 'From_' lines in mbox format files (look for the RPATHS #define).
Perhaps we should fix that too... [why do we use Return-Path: and
Delivery-Date: for this rather than X-MH-Envelope-From and
X-MH-Delivery-Date or something similarly unambiguous, anybody know?]

-- PMM