[6653] in SIPB bug reports
pmail / vi on Solaris
daemon@ATHENA.MIT.EDU (Jacob Morzinski)
Thu Jan 15 21:44:10 1998
To: bug-sipb@MIT.EDU
Date: Thu, 15 Jan 1998 21:43:53 EST
From: "Jacob Morzinski" <jmorzins@MIT.EDU>
I asked a friend why she continues logging into dec.dialup,
rather than the Solaris machines. She replied that it's because
of the poor behavior of pmail on the new dialups:
> lets say you do pmail,
> then you want to respond to a message, or mail a new one.
> Tnen you type some stuff and decide to edit it (I don't
> think this happens when you edit blank messages),
> when I go into vi, write stuff, and :w it, it think it's
> writing it fine, then I :wq, and it says "Fatal error in process"
> and sets the text back to what it was before I edited it
I tested vi for a while. The problem is that pmail checks the
exit status of vi, and Solaris vi returns a non-zero status very
easily. (The most common way to get one is to try a :s/foo/bar/
where there is no match for "foo". Each :s command like this
increments the exit status by one.) Even when the file was saved
successfully, the non-zero status causes pmail to discard the
result of the edits:
~
~
~
"/tmp/Re023149" 1 line, 3 characters
Fatal error in process.
(continue)
~q
Perhaps we should clobber pmail, and force it to stop checking
the return status of editors?
-Jacob
(MH's comp/whatnow has the same problem with Solaris's vi.)
