`conditional` type inference broken with `R.conditional.defaultCase()`

[stychu]

Hi, Im trying to get a grasp of functional programming but having some issues.

How I would pipe a value where it can be an object or undefined and then reference the object propertioes. I cant manage to narrow the correct types.

// value = {    groups: {value: string; initial: string}[]; } | undefined

R.pipe(
          value,
          R.conditional(
           [R.isTruthy, (name) => `Hello ${name}`], // This is unknow
            [R.isObjectType, (name) => `Hello ${name}`], // This is object
            [R.isDefined, (name) => `Hello ${name}`], // this is {}
            R.conditional.defaultCase(),
          ),
        );

Ideally I want an undefined if value is undefined or process the value with some more functions and return it

[eranhirsch]

It might not be something you are doing wrong; there are some known bugs with the typing of conditional, and it isn't working as well as we would want it to be. My recommendation is not to rely on it.

[stychu]

So there is no straight forward way to handle null | undefined values currently?

[cjquines]

this works fine if you pass a function to R.conditional.defaultCase, see playground

i'm surprised it is not a type error to call it without any arguments… i think we should change this?

[eranhirsch]

@cjquines, what is the error? defaultCase without any params would return undefined and is a shorthand to be used when migrating from Lodash where the default fallback returns undefined.

R.conditional.defaultCase() === R.conditional.defaultCase(constant(undefined));

[cjquines]

hm let me clarify. i think type inference is broken when called with () vs when called with (constant(undefined)). i don't know what we can do about this, but i'll keep this issue open for now and clarify the title