sed vs gnu sed

Vincent Hoffman vince at
Wed Nov 9 10:30:32 UTC 2011

'Hi all,
        I'm trying to move a script from a linux box to a freebsd box.
All going well as its just a bash script and bash is bash, however there
is one line I'm unable to use directly, as bsd sed (correctly according
to SUS at least, I believe[1]) appends a newline when writing to
standard out, gnu sed doesnt. example
[backup at banshee ~]$ echo -n "/boot:7:1:5; /:7:1:5; /var:7:1:5"  | sed -n
[backup at banshee ~]$


[backup at amber ~]$ echo -n "/boot:7:1:5; /:7:1:5; /var:7:1:5"  | sed
/boot:7:1:5;/:7:1:5;/var:7:1:5[backup at amber ~]$

is there any easy way to make our sed do the same as gnu sed here? 
for now I have encapsulated the whole thing in a subshell
[backup at banshee ~]$ echo -n $(echo -n "/boot:7:1:5; /:7:1:5;
/var:7:1:5"  | sed -n 's/[[:space:]]*;[[:space:]]*/;/gp')
/boot:7:1:5;/:7:1:5;/var:7:1:5[backup at banshee ~]$

Which works but seems a little hackish.


' Whenever the pattern space is written to standard output or a named
file, /sed/ will immediately follow it with a newline character. "

More information about the freebsd-questions mailing list