Path: utzoo!utgpu!news-server.csri.toronto.edu!bonnie.concordia.ca!uunet!zaphod.mps.ohio-state.edu!mips!pacbell.com!att!att!cbnewsl!bonnie!rbr
From: rbr@bonnie.ATT.COM (228-4197,ATTT)
Newsgroups: comp.unix.shell
Subject: Re: sed to remove newlines?
Keywords: sed
Message-ID: <1991May31.163100.20328@cbnewsl.att.com>
Date: 31 May 91 16:31:00 GMT
References: <6356@iron6.UUCP>
Sender: news@cbnewsl.att.com (NetNews Administrator)
Reply-To: rbr@bonnie.ATT.COM (Bob Rager)
Distribution: comp.unix.shell
Organization: AT&T Bell Laboratories
Lines: 34

In article <6356@iron6.UUCP> yeates@motcid.UUCP (Tony J Yeates) writes:
>Is it possible to remove newlines (i.e. \n) from a file using sed?  If
>so how?  We have tried various things without success:
>
>escaping \n -> \\n
>double escaping \n -> \\\n
>using \012
>
>but with no success.  Our man page incs. the following:-
>
>"\n   Matches a NEWLINE embedded in the pattern space. "
>
>If this can only be used for pattern matching but not exchange, it would seem
>to offer nothing over the use of $.
>
>[Our current requirement is to concatenate any lines ending with a ':' to  the
>next line in a file.   Although I can think of numerous other uses ( we
>usually end up using tr '\012' '<some char.>' as a work around).]

On input "sed(1)" converts the lint to a string. Therefore, the edit
functions never see the "\n" newline character. I think you may have
the wrong focus to solve the problem, think joining selected lines,
not eliminating newlines.

Why not use "ed(1)" as:

cp $1 $2
ed - $2 \\E-O-C
g/:$/j
w
q
E-O-C

Bob Rager