*.NonEmpty.Replicate

[cmorley191]

Please add the following function to the nonempty collection modules (not just List)

List.NonEmpty.replicate (count: PositiveInt) -> (value: 'a) -> NonEmptyList<'a>

[jimboreilly]

Isn't this just Seq.take : int -> seq<'T> -> seq<'T>

[ntwilson]

Actually, I think it's

let List.NonEmpty.replicate (PositiveInt count) value = 
  List.NonEmpty.create value (List.replicate (count -1) value)

[jimboreilly]

Ok I was misunderstanding the method signature, I read it as a function to copy (misinterpreted "replicate") the first n values

[cmorley191]

I was planning on adding a version of this method for the base modules as well, not just the NonEmpty modules. (the NonEmpty modules take a PositiveInt, so the base modules take a NaturalInt).

If I call it List.replicate, then any user opening SafetyFirst will have List.replicate shadowed by the function requiring a NaturalInt instead of an int. I imagine we don't want to break code just by opening SafetyFirst.

#44 has a similar situation where the type signature of groupBy should be changed by SafetyFirst, but we'd rather not override the function, and we want to reserve appending ' to the function name for functions that can fail (return Result).

The idea suggested over there was to use a different (but close enough in intellisense) name for the function. Do we want to continue with that? How about List.repeat? (edit: LazyList.repeat exists: infinite repetition. repeatN? repeatFinite?) List.reproduce?

We may want to consider if it would be annoying to have to guess whether the original or the "safe-signature" version is being used (e.g., "group" or "groupBy"?, "replicate" or "repeat"?, "init" or "initialize"?). I've gone through my keyboard and it looks like ' and _ are the only symbols we can stick on the end of an identifier.