From mboxrd@z Thu Jan  1 00:00:00 1970
To: 9fans@9fans.net
Subject: Re: [9fans] another dusty upas question
From: "Russ Cox" <rsc@swtch.com>
Date: Fri,  2 May 2008 13:30:43 -0400
In-Reply-To: <c9cc29ac8dfea8cfeeb3eb53669098c2@quanstro.net>
MIME-Version: 1.0
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 8bit
Message-Id: <20080502173332.7497C1E8C3A@holo.morphisms.net>
Topicbox-Message-UUID: 9d7e31e0-ead3-11e9-9d60-3106f5b1d025

> ☺.  i found it by reading the source.  i was trying to track
> down all the places where mailboxes are fidded in the upas
> source.
>
> it does seem very wierd and marshal does a shimmy to
> pull it off.  it would seem the only way to file outgoing
> & incoming mail to/from the same person in the same
> file.

ah, so the nedmail rf command invokes marshal -F.
i would be inclined to change -F to take the name
of the file, so that nedmail can be the sole arbiter of
what goes where.

> (offtopic: there are at least a few other crunchy bits.
> several copies of cistrn?cmp by various names.  there's

these could go.  they were in upas before they were in libc.

> also a reference to an rmail program in send/main.c
> does it (did it) exist on plan 9?)

i don't believe rmail ever existed on plan 9, but of course the
sources once compiled on unix too.  my understanding (which
may well be incomplete or incorrect) is that for much of the 1990s,
the bell labs mail gateways ran upas on irix machines,
and the binaries for those machines were generated by
compiling upas on plan 9, using vc and vl, but linking with
irix system call stubs.  perhaps behaving as "rmail" was still
necessary for that.  i also believe that's why vl has support for
writing elf binaries.  (but i might easily have the details wrong.)

russ