One more note - sort, uniq and merging to the string

It is just one more note to myself. We have list of string - for example like this:

bg, cal, cat, cd, chmod, clear, cmp, cp, date, df, diff, du, echo, exit, fg, file, find, grep,
groups, gzip, head, history, id, info, jobs, kill, ln, locate, ls, man , mc, mkdir, more, mv, 
passwd, ps, pwd, rm, rmdir, scp, sed, set, sleep, slogin, sort, ssh, tail, tar, touch, uname,
vi, wc, which, whoami.

I need the same list without repeating of items and sorted.

echo "bg, cal, cat, cd, chmod, clear, cmp, cp, date, df, diff, du, echo, exit, fg, file, find, grep,
groups, gzip, head, history, id, info, jobs, kill, ln, locate, ls, man , mc, mkdir, more, mv,
passwd, ps, pwd, rm, rmdir, scp, sed, set, sleep, slogin, sort, ssh, tail, tar, touch, uname, vi,
wc, which, whoami, , " | sed "s| *, *|\n|g" | grep -v "^ *$" | sort --unique |
paste --delimiters=, --serial | sed "s|,|, |g"

Special attention should be paid to paste - I always forget about this command :)
Actually, the last part can be rewritten with sed only and without paste:

... | sort --unique | sed ':a;N;s/\n/, /;ta'

lets look at the sed program:

#define label named 'a'
:a
#append to pattern space next line
N
#replace new line symbol by ', '
s/\n/, /
#and if replace has been successfully done go to to the label 'a'
ta
#in case last string nothing will be replaced and script will be finished

if we want to optimize replace statement we can use more complicated sed script:

... | sort --unique | sed -e ':x;$by;N;bx' -e ':y;s/\n/, /g'

in details:

{
#declare label 'x'
    :x
#if it is last line then go to label 'y'
    $by
#append to pattern space next line
    N
#go to label 'x'
    bx
}
#declare label 'y'
:y
#replace all new line symbol by ', ' in one pass
s/\n/, /g