Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!gem.mps.ohio-state.edu!rpi!tale
From: tale@pawl.rpi.edu (David C Lawrence)
Newsgroups: comp.unix.questions
Subject: Re: Import variables in to awk.
Message-ID: <1989Nov15.235518.8933@rpi.edu>
Date: 15 Nov 89 23:55:18 GMT
References: <10531@thorin.cs.unc.edu> <15919@bloom-beacon.MIT.EDU>
	<20774@mimsy.umd.edu>
Distribution: na
Organization: Rensselaer Polytechnic Institute, Troy NY
Lines: 29

In <10531@thorin.cs.unc.edu> warner@unc.cs.unc.edu (Byron Warner) writes:
Byron>  [file foo]
Byron> { print import,$0 }
Byron>  [command]
Byron> awk -F: -f foo /etc/passwd import='hello
Byron> why do I get just a list of logins?

Because the variable assignment has to come before file name.  I'm
also assuming here that the ' is a typo, or the absence of a match is;
either way variable assignment comes before the file list.  If you
change it to "awk -F: -f foo import=hello /etc/passwd" it will work.

This applies to V7 awk, nawk and gawk.

In <20774@mimsy.umd.edu> chris@mimsy.umd.edu (Chris Torek) writes:
Chris> All of the above is with respect to the 4.3BSD flavour of `awk'.  The
Chris> new awk (as described in the awk book) appears to open the first `file'
Chris> before executing the BEGIN statement, so that any assignments that
Chris> appear before the first real file happen before the BEGIN.  What GNU
Chris> awk does, I do not know (but the above technique will tell you).

Variables set as above are not available in the BEGIN block with gawk,
but a special option, -v, is provided to do this.  -v VAR=VAL will
assign VAL to VAR before script execution begins; another -v must be
specified for each variable you want to declare this way.

Dave
-- 
 (setq mail '("tale@pawl.rpi.edu" "tale@ai.mit.edu" "tale@rpitsmts.bitnet"))