Oops, that still leaves the issue of ~~. Well, seeing as it's a broken experimental feature, best if you simply avoid it!

Not only that, your smart-match version is far less efficient. Use one of the following:

# Unordered
my %diff; @diff{ @insttab } = ();
delete @diff{ @inst };
my @diff = keys %diff;
[download]

or

# Ordered
my %in_inst; @in_inst{ @inst } = ();
my @diff = grep { !exists($in_inst{$_}) } @insttab;
[download]

Both are O(N+M) (as opposed to the O(N*M) smartmatching approach).

---

Generally speaking, you'd want two different modules with the same interface in that situation. Simply load the correct module using if (the module).