Shell Programming and Scripting

Try it with the xpg4 sed.

SunOS: And with sed under /usr/xpg4/bin/sed ?

---------- Post updated at 11:47 AM ---------- Previous update was at 11:38 AM ----------

In fact, under SunOS, the syntax is:

Code:

sed -n '/client508/{
p
n
p
n
p
}' file

Only one command by line.
Regards.

Hi.
I sometimes use SunOS. I don't recall doing anything special when I installed it, and I have this:

Code:

OS, ker|rel, machine: SunOS, 5.10, i86pc
Distribution        : Solaris 10 10/08 s10x_u6wos_07b X86
ggrep grep (GNU grep) 2.5

So you might want to try ggrep. It is found in:

Code:

$ which ggrep
/usr/sfw/bin/ggrep

Best wishes ... cheers, drl

How about good ole ex... ex -s +'/^client508/;+2 p | q!' file

I don't use SunOS, but you are almost certainly mistaken. Both commands will fail on many sed implementations (not just SunOS) because POSIX requires a semicolon before the closing }.

While we're here, POSIX does not allow a semicolon after } for further commands. For portability, you'd have to use multiple -e options.

Both of those behaviors, omitting the semicolon before the closing brace and allowing a semicolon to follow a closing brace, are extensions to POSIX on which you cannot depend if portability is important.


Also, in case it isn't intentional, sed -n '/client508/{n;n;p}' won't print the line that matches the pattern nor the line that follows the pattern.

Regards,
Alister

The man posix of sed does not say quite right:
sed
For me, the best syntaxe for sed script is one command by line because all command support <newline> separator but many command do not support semicolon separator (example s command with w command).
And for <right-brace> as say the man:
Regards.

I know what POSIX says, but FWIW on Solaris this works:

Code:

sed -n '/client508/{p;n;p;n;p;}' file

as alister suggested...

and so does:

Code:

sed -n '/client508/{N;N;p;}' file

(also /usr/xpg4/bin/sed and also sed on HPUX and AIX BTW)

and probably also:

Code:

/usr/xpg4/bin/awk '/client508/' RS= file