vi(1) bug - defunct processes

[Vernon Schryver]

In article <14562 at bfmny0.UUCP>, tneff at bfmny0.UUCP (Tom Neff) writes:
> To heck with mailx(1), this is worse.

Ha!  Nothing is worse than mailx(1), unless you like gratuitous,
egregiously wrong features and stupid new bugs.  I wish BSD would (could?)
release 4.3BSD Mail source as they have done with the network code.

> 	AT&T UNIX System V/386 3.2.1
> [describes vi's failure to reap its dead children after ! operator]
>           The processes
> cannot be killed while 'vi' is still running, even with signal 9 from
> root.  Only exiting vi fixes it.
> "We walked on the moon --	((	Tom Neff
> 	you be polite"		 )) 	tneff at bfmny0.UU.NET

In ISC 2.0.2 doing something with :! such as  ":!sh" or ":!ls" seems to
cause vi to do the necessary wait(2)'s.  That is, after "!" becomes dead,
":!ls" makes everything wonderful again.

The bug was in Microport 2.2-3.0e, and is in the SVR3.2 source tape, so
it is not surprising to find it in AT&T's retail product.

Vernon Schryver
vjs at sgi.com