You are not logged in.
- Topics: Active | Unanswered
#1 2010-10-01 01:09:58
- Rad3k
- Member
- From: Poland
- Registered: 2009-05-09
- Posts: 36
[SOLVED] Weird way of testing for null string in sh.
It may be a dumb question, but I couldn't find the answer. Are there any reasons to use
if [ "x${string}" != "x" ]instead of
if [ -n "${string}" ]or just
if [ "${string}" ]?
As for the first syntax, I can see some disadvantages (it's longer, less readable, probably also less efficient) while no obvious advantages over the other ones. Is the first version more portable, or something like that?
I ask because I'm writing my own encryption hook for mkinitcpio (to add support for resuming from encrypted swap), and I've encountered this syntax in existing code. Now I wonder whether it's safe to always go the usual way in this hook... 
Solved, see hbekel's answer
Last edited by Rad3k (2010-10-01 13:53:19)
Offline
#3 2010-10-01 06:03:48
- hbekel
- Member
- Registered: 2008-10-04
- Posts: 311
Re: [SOLVED] Weird way of testing for null string in sh.
"The x"$foo" hack is required for code that must run on ancient shells which lack [[, and have a more primitive [, which gets confused if $foo begins with a -."
Offline
#4 2010-10-01 13:50:55
- Rad3k
- Member
- From: Poland
- Registered: 2009-05-09
- Posts: 36
Re: [SOLVED] Weird way of testing for null string in sh.
Thanks hbekel 
So portability (probably taken to the extreme) is the reason here. I wonder whether busybox's ash qualifies as such an ancient shell. Will have to check it. Marking as solved.
Offline
#5 2010-10-01 13:55:27
- karol
- Archivist
- Registered: 2009-05-06
- Posts: 25,440
Re: [SOLVED] Weird way of testing for null string in sh.
It's easier to port a shell than a shell script. -- Larry Wall
Offline
#6 2010-10-01 21:51:48
- falconindy
- Developer
- From: New York, USA
- Registered: 2009-10-22
- Posts: 4,111
- Website
Offline