5

Recently I was working on performing a sed operation on a config file. I found that the sed command is behaving differently with the parameter -ir vs -ri :- 

[root@node system]# sed -ri 's|(^[[:space:]]+[Kk]ernel.*$)|\1 transparent_hugepage=never|' temp_file
[root@node system]# echo $?
0

[root@node system]# sed -ir 's|(^[[:space:]]+[Kk]ernel.*$)|\1 transparent_hugepage=never|' temp_file
sed: -e expression #1, char 60: invalid reference \1 on `s' command's RHS
[root@node system]# echo $?
1
Improve this question

1 Answer 1

Reset to default
11

The -i option to sed takes an option argument, a filename suffix to use for the backup file when editing the input file in place. With GNU sed, this option argument is optional, but will obviously be used if supplied.

Using -ir tells sed that the filename of the backup file should have the original name of the input file suffixed by the character r. Since -r is now not used, the back reference in the replacement part of the expression is no longer recognised as a valid, since there is no \( ... \) group in the pattern (which is how you capture a part of a pattern with a basic regular expression).

In conclusion, sed -ri is not the same as sed -ir.

Improve this answer
2
  • Yes I checked the system also, Thanks for your help Commented Aug 9, 2017 at 11:08
  • n.b. the argument is required on BSD seds Commented Aug 9, 2017 at 17:59

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.