Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!uunet!mcsun!hp4nl!star.cs.vu.nl!maart
From: maart@cs.vu.nl (Maarten Litmaath)
Newsgroups: comp.unix.questions
Subject: Re: What's wrong with this Bourne shell script?
Message-ID: <7248@star.cs.vu.nl>
Date: 6 Aug 90 16:15:49 GMT
References: <1990Aug3.193231.3166@silma.com> <3821@auspex.auspex.com>
Sender: news@cs.vu.nl
Reply-To: maart@cs.vu.nl (Maarten Litmaath)
Organization: VU Dept. of Computer Science, Amsterdam, The Netherlands
Lines: 35

In article <3821@auspex.auspex.com>,
	guy@auspex.auspex.com (Guy Harris) writes:
)...
)The Bourne shell runs some shell constructs in a sub-shell; apparently,
)the construct you gave is one of them.  [...]

[The construct was ``command | while read line; do something; done''.]

)This may change when shells become compliant with POSIX 1003.2, which
)will presumably happen sometime after POSIX 1003.2 leaves the draft
)stage and is approved; 1003.2 may put stricter controls on when a shell
)may do something inside a subshell.  I don't have recent drafts handy;
)Maarten?

I haven't read Draft 10 yet, but I pleaded for the `while' loop from the
example to be run in the current shell instead of a subshell (if the user
wants a subshell he can always use parentheses).  The advantages are
obvious.

In any case, _redirected_ loops already must be run in the current shell,
notwithstanding the current practice in a lot of Bourne shell implementations
to create a subshell:

	var=before
	while read line
	do
		var=after
	done < file
	echo var=$var

This example will print `after' on POSIX 1003.2 conformant systems,
`before' on many current systems.
--
   "UNIX was never designed to keep people from doing stupid things, because
    that policy would also keep them from doing clever things."  (Doug Gwyn)