Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uunet!decwrl!shelby!morrow.stanford.edu!lindy.stanford.edu!whenry
From: whenry@lindy.stanford.edu (homo obsolescensis)
Newsgroups: comp.lang.perl
Subject: Re: Novice Question: How is -e supposed to work?
Message-ID: <1990Aug16.172303.12982@morrow.stanford.edu>
Date: 16 Aug 90 17:23:03 GMT
References: <1990Aug16.070010.26529@morrow.stanford.edu> <11415@ogicse.ogi.edu>
Sender: news@morrow.stanford.edu (USENET News System)
Organization: Stanford University Libraries
Lines: 36

In article <> schaefer@ogicse.ogi.edu (Barton E. Schaefer) writes:
>In article <> whenry@lindy.stanford.edu (homo obsolescensis) writes:
>} 
>} I am having trouble using the -e switch.  the man
>} pages give the following example: (lets call this file "go")
>} 
>} 	#!/usr/bin/perl -pi.bak
>} 	s/foo/bar/;
>} 
>I'm curious what this has to do with the -e switch (as mentioned in
>the subject header).  

Urgh, am I embarassed.  my question is about -i, not -e.  just a typo,
caused by lack of sleep.  the above fragment is given in the man pages 
as an equivalent to 

	perl -p -i.bak -e "s/foo/bar/;" ...

Mr. Schaefer suggests

>	perl -pi.bak -e 's/foo/bar/;' < bas
>
>and everything will be fine.

This does indeed work as far as -e is concerned (s/foo/bar is done)
but not as far as -i (or, more likely I dont understand -i).  shouldnt
your fragment result in bas being overwritten with the changed text?
and the original bas renamed to bas.bak?
when I do what you suggest bas is just written to the screen (with, of
course, the changed text).  the man pages give yet a third
"equivalent" example (much longer, so I wont quote it), in which
bas is indeed overwritten (and a .bak made)

(Im really sorry about the confusion caused by my typo and much
appreciate your answer, which I learned something useful anyway).
Walter Henry