Question

if a file contains multiple columns, seperated by comma, like this:

aaa,1,4,4,5,7
bbb,1,4,9,1,2

Is there difference between ‘sort -t, -k1 file.txt’ and ‘sort -t, -k1,1 file.txt’?
though with the example above, there is no difference, but in some of my project case, it
does has difference, but the difference is reflected in the case that I use the sorted file to join, and the join command throws out exception that ‘join: file 2 is not in sorted order’ (at that time , I use ‘sort -t, -k1 file.txt’) . Later I use ‘sort -t, -k1,1 file.txt’, and join command works well then. Can anybody tell me why?

Answer 1

sort -k1 means sort starting on key 1 till the end of the line. sort -k1,1 means sort from key 1 to key 1 (so only the first key). On my machine, the two make a difference if I specify stable sort with -s:

~ $ cat test.txt
aaa,1,4,4,5,7
aaa,1,3,9,1,2

~ $ sort -t, -k1 -s test.txt
aaa,1,3,9,1,2
aaa,1,4,4,5,7

~ $ sort -t, -k1,1 -s test.txt
aaa,1,4,4,5,7
aaa,1,3,9,1,2