Re: [tlug] sed or awk or what?

[Sven Simon <sven@example.com>]

Quoting Bruno Raoult <br@example.com>:

> Or shorter:
> 
> sed -e '2s/\(.* \)\{8\}[^ ]*/\170' < A.txt
> 
> But I did not test on other systems than mine (linux):
> I am always afraid when I use these kind of regexp
> with sed, I had some bad experiences at work :-(
> 
> I am not sure this works on Solaris/AIX/MacOS X (yes,
> the new one we can include in Unix tests!) and others. 
> 
> One example: some sed versions will need a "/" at the
> end of the substitute command: It would be better to
> add it in all cases.
> 
> br.
> 
> On Tue, 2002-08-06 at 21:07, Bruno Raoult wrote:
> > What about this one?
> > 
> > sed -e '2 s/\(.* \)\{8\}[^ ]*\(.*\)/\170\2/' < A.txt
> > 
> > br.
> > 

Thanks Bruno, you've pointed me in the right direction.
But after digging up the sed section in good old "Unix Power Tools"
I saw we can set a numeric flag for pattern matches in the substitute
command, so I came up with this:

sed -e '2s/[^ ]* /70 /9' A.txt

Easy enough, no? It sure took me a while, though. 
Ah, Regular Expressions - what a headache!