I avoided associative arrays in my suggestion, because I only have access to bash version 3.2.57(1)-release (x86_64-apple-darwin16) and it doesn't have associative arrays. I normally use ksh instead of bash (and it has had associative arrays since 1993). Unfortunately, bash and ksh declare associative arrays incompatibly. As you have shown above, bash declares an associative array with:

Code:

declare -A arrayname...

while ksh uses:

Code:

typeset -A arrayname...

with that one line change to your code, it will work with a 1993 or later ksh instead of a version 4 or later bash.

I agree that using an associative array is the easy way to avoid duplicated values when creating an array. But note that you when you say that it "automatically sort out duplicates", the sort order produced by ${!arrayname[@]} is an alphanumeric sort of the indices; not a numeric sort. So if the values provided as the array indices are 5, 20, and 300; the output from:

Code:

echo ${!arrayname[@]}

will be:

Code:

20 300 5

not the:

Code:

5 20 300

that would be produced by a numeric sort.

The OP for this thread never said whether or not maintaining the order of the input values was desirable or important; nor whether sorted output would be a nicety or a nuisance. If numerically sorted output is important, the user entering values would either need to supply leading zeros so all numeric value are the same width, or the list of values will have to be sorted after duplicates are removed (either with an associative array or with the indexed array used in the original post in this thread).