UNIX for Advanced & Expert Users

Hi,

The OS I am using is SunOS 5.9 as seen by the command "uname -a". My problem regarding sort is that solaris sort does not have the -s option which I need exactly.

Suppose, the input is like this:

345 t
123 o
567 r
345 a
345 c

I want to sort only on first field and if the field is equal, I want the lines to appear as they are in the input. I tried this:

sort -k 1,1 filename

The output was:

123 o
345 a
345 c
345 t
567 r

But I want the output to be like this:

123 o
345 t
345 a
345 c
567 r

I know that this can be achieved if -s option could be given. But my sort doesn't have it. Any help would be greatly appreciated. Thanks.

Try this. Number the records, sort, then remove the record numbers. Untested on Solaris.
cat -n filename | sort -k 2,2 | cut -f2-

Thanks for your suggestion, but it does not work. Actually I tested it on cygwin also which has different sort (GNU sort) than solaris sort and it does not work on both of them.

As far as I understand what sort does is this: when it finds the field to sort is equal on some lines, it tries to sort the lines completely as if there were no field specified. I just want to stop this behavior. Maybe some tricks with -k option could do it, I don't know. I haven't found out yet.

methyl's solution works for me on Solaris:

Code:

$ cat file
345 t
123 o
567 r
345 a
345 c
$ nl file|sort -k2,2n|cut -f2- 
123 o
345 t
345 a
345 c
567 r

> cat file35
345 t
123 o
567 r
345 a
345 c
> cat -n file35 | sort +1.1 -1.2 | awk '{print $2" "$3}'
123 o
345 t
345 a
345 c
567 r

At last sorted it out..

methyl's solution along with -k 2,2n option worked as shown by radoulov. The n option is important.

Many thanks for the solution.