Sed + delete line only if line ended with specific character

Solaris generally doesn't have the GNU version of anything by default. This means that the options and arguments you supply will often have to be different to get the same behaviour out of utilities like sed.

Sometimes it will not be possible to get the same behaviour at all, for instance, sed -i won't work on Solaris unless you have installed the GNU version of sed.

From memory, the sed on Solaris supports the "basic" set of regular expressions and this doesn't include the + modifier. You can simulate a + like this:

sed 's/[^0-9][^0-9]*//'

Or you can just use a * for this case:

sed 's/[^0-9]*//'

Solaris usually also has the BSD versions of these utilities installed under /usr/ucb. they are often very similar but occasionally have important differences, such as with ps.

 echo "121.122.121.111] other characters in logs from sendmail...." | sed 's/].*//' 

So if you have a file full of lines like that you can do

 sed 's/].*//' filename