Deprecate fmap(walk, f, x, ys...) for execute(walk, x, ys...)

[gaurav-arya]

Deprecates fmap(walk, f, x, ys...) for runwalk(walk, x, ys...), with no higher ambitions. Resolves #62

Edit: runwalk now renamed to execute

[gaurav-arya]

Closing per discussion in #62

[gaurav-arya]

Well, reopening in case the little bit of work I did with the deprecation becomes something we want to use in the future, but this PR isn't ready for merge.

[darsnack]

I think we should go ahead and do this, but use traverse and note in the docstring that this is post-order traversal.

Just getting rid of the confusing fmap(walk, f, ...) method will be an improvement even if the naming isn't perfect. I imagine a proper renaming that mirrors prior literature on "reduce," "traverse," "fold," "map," etc. will be bigger, breaking overhaul anyways. And so the name we choose here won't make it significantly more painful that it is going to be anyways.

[gaurav-arya]

Sounds good! Happy to do a renaming. I'm musing about whether traverse is the appropriate name, though. I don't speak Haskell, but from https://stackoverflow.com/questions/7460809/can-someone-explain-the-traverse-function-in-haskell it looks like it's an fmap that guarantees a particular ordering and allows effectful functions. In particular, it seems like it should definitely have the same signature as fmap (and the stack overflow answer notes that the distinction between fmap and traverse is blurred in an impure language.)

So maybe traverse isn't the best name for this function? To me, it feels like this a lower-level primitive which is very much to do with the specific concept of a walk in this library, So maybe something like dowalk to avoid the oxymoron? Very happy to be contradicted and for other suggestions in the name.

Edit: I guess your point probably is that we should just use traverse because it's a fitting name, and ignore any FP comparisons. I mostly agree, but I wonder if it's still better to avoid using that specific name which already has other implications, to avoid any confusion

[darsnack]

How about descend(walk, x, ys...) (or execute or enter)?

But yes, my point was that the closest FP concept to what's going on is a post-order traversal, so we should just use that.

[CarloLucibello]

@gaurav-arya let's decide a name and merge this, if you have some time

[gaurav-arya]

Sure -- have gone with execute

[darsnack]

Thank you @gaurav-arya for all the hard work and thoughts here!