`Supply.map` should respect the `Slip` semantics as `List.map` does

[FCO]

https://colabti.org/irclogger/irclogger_log/raku?date=2019-10-21#l133

[lizmat]

Feels this is actually a bug. We should expect the same semantics of .map regardless on which type it is being called.

[jnthn]

We should expect the same semantics of .map regardless on which type it is being called.

Well, yes and no: there's at least 3 implementations of map which can largely be said to "do the same thing" (iterable, parallel iterable, reactive), but equally they all have their own semantic details due to the nature of the processing model they're working with. (For example, last has much more "interesting" semantics on a parallel iterable.)

I don't think we can consider map in isolation when it comes to how Slip interacts with Supply; probably this comes up in other places too, and we'd want to make a consistent decision across them all.

[jnthn]

Ah, the other thing I wanted to point out: consider .flat on Supply, which actually is about flattening out nested Supply instances, but does not touch Lists or Arrays emitted on them. Likewise, we'd not expect .flat on a List that contains a Supply to treat that Supply as anything but data. (Effectively, all these monadic constructs know what is their structure and what is their value.) So the question really boils down to whether we consider Slip to be a concept distinct to the Iterable model, or if we thing it's more cross-cutting than that.