I am using egrep to list a bunch of lines of code from various cpp and h files.
I want my list to be primarily alphabetically sorted on the name of the file (so the .h file follws its .cpp file) and then numerically by line number.
I am using:
egrep -i -n -T '<str1>|<str2>' *.cpp *.h | sort -t: -k 1 -k 2.1,2.3n
I have tried all sorts of "k" combinations but I always get the files sorted as I want, but the line numbers are alphabetc ie something like 1 10 13 18 2 21 3 etc.
If I put the line numbers as the first key field they come out in the correct order.
I tried it by using cat on the password file and passing it to sort and got similar problems with second field.
I checked various sources and I think it should work.
Any ideas?

Try using sort's -g, -n, or -h options.

Have tried many variations to -k option. The examples imply that you can sort on more than one key, but I am not convinced!
Here is a little experiment:
List of test file:
$> cat test.dat
aaaa 2
aaaa 4
aaaa 21
aaaa 12
aaaa 1002
bbbb 30
bbbb 3
bbbb 20
Sort by second numeric field as primary - works OK
$> cat test.dat | sort -k 2n
aaaa 2
bbbb 3
aaaa 4
aaaa 12
bbbb 20
aaaa 21
bbbb 30
aaaa 1002
Sort by first alpha field as primary then by numerical field - fails
$> cat test.dat | sort -k 1 -k 2n
aaaa 1002
aaaa 12
aaaa 2
aaaa 21
aaaa 4
bbbb 20
bbbb 3
bbbb 30

Ah, ok. I see what you mean. That was a tricky one. I had to go to the sort info page to figure it out.
When sorting by multiple columns, using only -k1 messes it up because that sorts the value from the beginning of the field to the end of the line. You have to tell it to focus only on the current field.

Note that in this example the full -k2,2 is not really necessary, since there are no following fields after it that need to be ignored. Only the first field needs the full specification. But it would be needed if you wanted to further sort the data by a third field.

1 members found this post helpful.

Many thanks David the H. That works exactly how I wanted it to.