Subject: question about signals and system() and mutt
To: None <>
From: Ben Rosengart <>
List: tech-userlevel
Date: 07/09/2001 19:40:56
Hello NetBSDers,
  We are having a problem at Panix which we think we understand;
however, we are not sure what the correct solution is.  I am hoping
that someone here can enlighten me.

Here's the scenario: a user uses mutt to compose a message.  Mutt
spawns $EDITOR using system(3).  At some point during the composition
of the message, the user presses ^C, which is caught by the editor
and handled appropriately.  However, when the user quits the editor
in order to send their message, mutt reports something like "Error
running "vi '/tmp/mutt-panix1-28255-0'"!  This is only a problem
because the user thinks it is, and calls us.

We think that system() is returning non-zero because the shell is
terminating abnormally, and we think the shell is terminating
abnormally because it gets the SIGINT delivered to it as soon as
the editor terminates, but we're not sure.

My questions are this: given that a user ought to be able to send
SIGINT to their editor without unduly affecting the process that
spawned the editor, which piece of code is misbehaving (mutt,
system(), or sh) and what would be more correct behavior?

Thanks in advance.

Ben Rosengart     (212) 741-4400 x215