A.P. Lawrence

2005/05/26 sort -g vs. sort -n

When sorting numbers, you can use either -g or -n. I've noticed that Linux users without Unix experience tend to use -g, while old Unix folk are often unaware of that flag at all and continue to use -n. There is a difference between the two flags, although "man sort" doesn't explain it - "info sort" does a better job.

For ordinary numbers, -g and -n are identical, and although -g is slower, for small input sets it really doesn't matter. But when you mix in scientific notation, -n does not work:

$ cat t
 123
12
11
9
453
99
10e1
10e0
101
1
8.95
$ sort -n t
1
8.95
9
10e0
10e1
11
12
99
101
 123
453
$ sort -g t
1
8.95
9
10e0
11
12
99
10e1
101
 123
453

I mentioned that "sort -g" is slower. That's because it calls "strtod" to convert to double-precision floating point. When -n is used, sort simply aligns decimal points (real or assumed) and then does an ordinary string comparison. That's much faster, but of course it fails to handle scientific notation. The "faster" is, as noted, usually unimportant:

$ wc -l t
1321 t
$ time sort -g t > /dev/null

real    0m0.013s
user    0m0.012s
sys     0m0.002s
$ time sort -n t > /dev/null

real    0m0.009s
user    0m0.008s
sys     0m0.001s

You may have "-g" even if your man page doesn't mention it: my Mac OS X Tiger has -g but the man page doesn't mention it.

Comments /Words2005/2005_05_26.html


Add your comments

Why no "Digg this!" etc.?

Enter your email address for automatic notification of new posts here
(be sure to whitelist 'feedburner.com' if you use spam filtering)

Phone and Email Support at Monthly rates

Have you tried Searching this site?

Unix/Linux/Mac OS X support by phone, email or on-site: Support Rates

This is a Unix/Linux resource website. It contains technical articles about Unix, Linux and general computing related subjects, opinion, news, help files, how-to's, tutorials and more. We appreciate comments and article submissions.

Publishing your articles here