|
|
|
December 21st, 2007
 | 03:30 pm - Apparently there's a bug in sort ... or something.
So, a strange Unix thing I noticed today.
Given the following file :
> cat list2
1
1
2
A
B
Now try sorting it. First with just "-u" (sort uniquely) :
> sort -u list2
1
2
A
B
Now add "-n" to sort numerically (if the field looks like a number) :
> sort -un list2
A
1
2
Where did the "B" go ??
I've seen this in sort under Solaris 8, the GNU version in Ubuntu 7.04 and the gnuwin32 version on Windows. Surely it's a bug ?
Current Mood:  geeky
|
![[User Picture]](https://l-userpic.livejournal.com/104550077/566732) | | From: |  drplokta |
|---|
| Date: | December 21st, 2007 03:47 pm (UTC) |
|---|
| | | (Link) |
|
A and B have the same numeric value (i.e. not a number), so when you sort by unique numeric values you only see one of them. What he said!
"-n, --numeric
An initial numeric string, consisting of optional white space, optional sign, and a nonempty string of digits with optional decimal point, is sorted by value."
as A and B consist of a line with zero digits etc. they are the same "no numeric value".
Thanks Mike, that would have puzzled me for a while too, as I'd have assumed it would have dropped both the A and the B rather than just putting one up. Or evaluated them as Hex digits .... 0xA 0xB (which would have been silly) or by their Ascii value (even sillier!) Ah, that makes perfect logical sense. It's just slightly counter-intuitive at first glance :-)
It means "sort -un list1" produces different output to "sort -u list1 | sort -n". Hmmm ... wonder if any of my old scripts need to be revised to take this into account ?
Well that's good to be aware of :) |
|
|
|
|
LiveJournal.com |
