Well ok i’ll bite. (way off the OP post) I’m a big proponent of using the switches to be sure and accurate , but in the case of the output of du I can’t see where it makes a difference. The default is sorting each line as one whole field, and the output of du always starts with spaces and a number - so by definition that will always sort in an intuitive way - in fact I’ve not found an example where using nk1,1 changes the output in any way from just using sort (speaking on the output of du only).
The only difference I can find is a trumped up example whereby there is no numeric, then sort treats that as putting it after the numerics (any character > than any number) and -n treats any non-numeric starting value as 0 such as > cat test.txt | sort -1file 001 file 01 file 011 file 11 file 110 file 12 file file > cat test.txt | sort -nk1,1 -1file file 001 file 01 file 011 file 11 file 12 file 110 file But still that wouldn’t happen with du as a du ouptut as du will output minimum 0 for everything. So the switches are unneeded. du nonwithstanding - -n would i think be confusing in cases where, like the above, you have mixed beginning numeric and alpha, as I would think most would expect sorting of the numerics first, then the alphas, and not the non-intuitive sorting of “file” in between “-1file” and “001file” vs the former of having it after all the numerics. Last - I think -k1,1 is pretty much superfluous here given the space separator between the numeric size and folder/file name (sort assumes any combo of spaces tabs are a field seperator by default) - as I read the man page k[m.n,p.q] means sort starting at the qth character (here 0) of the mth field (hear first field) and continue through the qth character (again 0) of the pth field (here 1) - so k1,1 means the key is the entire first field - which would be the default if there was any space after it. if there is a tie, then it moves to the next character. Only in examples like the last one, would -k[1,3] change the order. > cat test.txt d0000002 file3 d0000001 file2 d0000003 file1 > cat test.txt | sort -nk1,1 d0000001 file2 d0000002 file3 d0000003 file1 > cat test.txt | sort -nk1 d0000001 file2 d0000002 file3 d0000003 file1 > cat test.txt | sort -nk1,3 d0000003 file1 d0000001 file2 d0000002 file3 I could see using something like: cat test.txt | sort -nk1.8 (note period not comma) y0000001 file2 x0000002 file3 z0000003 file1 to sort by the numeric at the end of the first field. But this again is a made up example not an output from du. But I digress … this was fun! … wait what was I talking about... Chad > On Feb 19, 2016, at 9:35 AM, Paul Gilmartin > <0000000433f07816-dmarc-requ...@listserv.ua.edu> wrote: > > On 2016-02-19, at 07:48, Bigendian Smalls wrote: > >> Tim have you tried this from a shell at the root of your ZFS partition (or >> just root / ) >> >> du -sk | sort >> > BTW, I'm more comfortable with: > > du -k | sort -nk1,1 > > -- gil > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
