`obj` position in argument order

[jariji]

Higher-order functions like map(f, obj) usually put the data argument last. In Accessors, it's usually hof(obj, optic, val) or modify(f, obj, optic) where the data argument comes earlier, which I didn't expect. I'm interested what explains this decision?

[jw3126]

In very early Setfield.jl commits the order was actually like you suggested. However:

• I found it confusing that the order is switched in @set obj.field = val and set(@lens(_.field), obj, val)
• There is one good reason to put functions first, namely do blocks. However, this advantage is lost if the first argument is an optic.
• Also back in the day I missed/did not take seriously enough the viewpoint, that a lens is just a function that has a set method as well.

[jariji]

This does bother me a bit. I always expect it to be

modify(cumsum, reverse, xs)

like all the other functional programming things and I have to check the docs to remember that the optic comes last and the data comes in the middle. I'm happy with the function coming first.

[aplavin]

I personally also find the current order totally reasonable. Anyway, there's probably no objective better/worse here, and it's not worth a breaking change.