Reverse key-filter arguments? about liquidata

[gonzaponte]

I think {key : predicate} is more intuitive as we read from left to right*. But in the end, this is the same dilemma as with get/put:
(get.a.b * f >> put.x.y is the most readable form as a whole (IMHO), while
(f * get.a.b >> put.x.y) makes more sense in terms of understanding what operations are applied to the data.

I think I prefer {key : predicate}, but I don't have a strong opinion about it.

[jacg]

this is the same dilemma as with get/put

get and put give the user the choice. For key-filters we have to pick one order to which the user will have to conform.

(get.a.b * f >> put.x.y is the most readable form as a whole (IMHO), while (f * get.a.b >> put.x.y) makes more sense in terms of understanding what operations are applied to the data.

Exactly, it's context-dependent, so it's good to leave the choice to the user.

I think I prefer {key : predicate}, but I don't have a strong opinion about it.

I prefer it too, but : key is essentially an option, and it seems weird to have the option precede the main item. Consider sorted(data, key=xxx) etc.

I think I'll probably change it to { key : predicate }, but I don't think there's any hurry, it's not as if we're going to get a thousand users by next week.