Very true, which is why I respectfully disagree with Dirk that all ml
list-like functions should live in a List class.
Not quite what I meant, sorry. I mean the relationship should be like the
standard string library: available as a dot-call for all tables (at the user's
risk that it makes sense) and as a colon-call for Lists.
BTW apart from preferring Sequence because it is decreed, I object to
List because it suggests O(1) insertions and deletions. I know the
term is used misguidodly in a language that seems to have influenced
the design of Microlight, but why propagate a mumpsimus?
BTW, #s is (a) unavoidably O(N) [1] and (b) only works for 5.2. So
perhaps we need count_keys() as well.