Can the sort command work from the right side?

lastcomm has some fields which are not always output, so it's much easier if I could sort from the right hand side, however after reading the manpage it does'nt appear possible.

Is there an easy shell solution?

lastcomm | egrep -v '0.[0-9][0-9]' | sort +1nr

> sort --v
sort (textutils) 2.0.21

Regards

Is this what you want:

sort +5 +4 +0 <somefile>

This wil sort coloumn 6, then 5, then 1 (in that order)

+4 (and +0) will only be used if:

1) +5 has similar entries (sub sort)
2) there is no column 6

You could use -t to set a seperator (standard space/tab).

I'm not familiar with lastcomm, so I don't know what kind of output it generates.

Nope this is'nt what I want because lastcomm has some fields that are not always output/displayed. Those fields are for

+ flags, as recorded by the system accounting routines:
S -- command executed by super-user
F -- command executed after a fork but without a following exec
C -- command run in PDP-11 compatibility mode (VAX only)
D -- command terminated with the generation of a core file
X -- command was terminated with the signal SIGTERM

Is it possible to sort from the right side using Unix?

Failing that, is it easy in perl?

Example output

[root@myhost]# lastcomm
myprog abc ?? 44.99 secs Mon Nov 10 14:39
perl S root ?? 2.64 secs Mon Nov 10 14:39
ld abc ?? 56.50 secs Mon Nov 10 14:37
gcc abc ?? 3.54 secs Mon Nov 10 14:37
gcc abc ?? 14.56 secs Mon Nov 10 14:36
gcc abc ?? 1.08 secs Mon Nov 10 14:36
perl S root stdin 2.40 secs Mon Nov 10 14:36

Regards

Changed your example output a bit. Timestamps/dates only, no layout stuff.
Looks like this
Sorted on the last field. Play with the sort options to make it the way you want.

Don't know if you see what it is the code is doing, so i'll explain just in case......

The dot (.) in the numbers before secs seem to be unique so I use that to split the line in two. The first field has the 'unpredictable' part, the second field has a layout that doesn't change and is needed for the sort.

I switch the fields. first the predictable field followed by the unpredictable field. I use a dot (.) as the ouput field (need that later on to switch the to fields back).

Now I can do a sort with predictable field(s).

After sorting we switch the 2 fields back to their original position and print the line.

I'd like to sort by process load :-)

The example is fine if the load is 0.xx, but if it's y.xx then these must be sorted and not just xx.

I'm sure this is easy, I just can't see it.

Thanks for your reply.

Regards

Mmmmmm.......... So you do not want to sort on the last field(s), but on the process load field. Why didn't you say so

Using the same input file as before:

$ cat end2front.input | sed 's/ \([0-9]\)/~\1/' | awk -F"~" 'BEGIN{OFS="~"}{ print $2,$1 }' | sort -n | awk -F"~" '{ print $2,$1 }'

gcc abc ?? 1.08 secs Mon Nov 10 14:34
perl S root stdin 2.40 secs Mon Nov 10 14:35
perl S root ?? 2.64 secs Mon Nov 10 14:38
gcc abc ?? 3.54 secs Mon Nov 10 14:37
gcc abc ?? 14.56 secs Mon Nov 10 14:36
myprog abc ?? 44.99 secs Mon Nov 9 14:39
ld abc ?? 56.50 secs Mon Nov 9 14:35

Thank you! Works great! Very useful indeed!