id and (.) are ambiguous about base-prelude HOT 5 CLOSED

BasePrelude is only supposed to be used in combination with the NoImplicitPrelude pragma.

from base-prelude.

I didn't realize that. It's a shame because hiding a function has to be done only once (by one people) whereas the pragma as to be done over and over.

from base-prelude.

The project is not about doing something once, it's about replacing the standard Prelude with a more general one. You'd discover way more functions in BasePrelude, which would conflict with the standard Prelude, if you did the change you suggested.

BTW, FYI, pragmas can be done project-wide, so member-hiding is not an issue. E.g.: https://github.com/nikita-volkov/hasql/blob/81f78acf0d90913360ec2d991b246daed948ca81/hasql.cabal#L44-L45

from base-prelude.

Fair enough, I just think it would be a "better" replacement if you could just do import BasePrelude and nothing else. You can indeed do pragmas project-wide, but you still have to do it for every project and every target. Anyway, that's your decision.

from base-prelude.

I just think it would be a "better" replacement if you could just do import BasePrelude and nothing else.

It wouldn't be a replacement then, but an extension. Either way, you'd have to lose an important property of BasePrelude: the preference of polymorphic functions, which would conflict with their monomorphic counterparts from Prelude. (.) and id are exactly a case of that, with the ones exported by BasePrelude being generalized to the Category typeclass.

You can indeed do pragmas project-wide, but you still have to do it for every project and every target.

And what's the problem about that?

from base-prelude.