InferType inconsistent with module implementation

[colinhacks]

If you know how to fix the issue, make a pull request instead.

• I tried using the latest stable version of tsc. https://www.npmjs.com/package/typescript
• I tried using the @types/xxxx package and had problems.
• I have a question that is inappropriate for StackOverflow. (Please ask any appropriate questions there).
• Mention the authors (see Definitions by: in index.d.ts) so they can respond.
  • Authors: @vtserman @MoretonBayRC @sseppola @YashdalfTheGray @robertbullen @sat0yu @dancrumb @deskoh @mauricedb @kalley

If you do not mention the authors the issue will be ignored.

---

I have a suggestion for an improvement to the typings. Happy to submit a PR but I wanted to make sure this change makes sense first.

Currently, the yup package treats all object properties as optional by default:

const schema =  yup.object({
  asdf: yup.string()
})
schema.validate({})  // passes

Yet yup.InferType infers the type as: { asdf: string } when really it should be { asdf?: string }.

The idea behind this change is to make all calls to yup.string, yup.number, and yup.boolean optional by default, which aligns with the behavior of the actual package implementation. I have this working on my local machine, but I wanted to run this by you guys in case there's a reason it's currently done this way.

[mauricedb]

Interesting idea but I am pretty sure this would break a lot of code. I did consider doing this when inplementing InferType<T> but decided against it.

For one the inrerred type would change all of a sudden when users update to the newer versions of the type definition making this a large breaking change. By itself not a reason not to do so, if something is broken it should be fixed.

However the semantics of a required property in Yup is not the same as in TypeScript. For example a Yup.array().of(Yup.string()).required() will fail validation if you pass in an empty array. A required array should not be empty: empty arrays are also considered 'missing' values.

In general there isn't a one to one match between Yup and TypeScript concepts so this is never going to be perfect. Another example of that is a Yup.date() that will pass validation if you use the string "2020-02-14T07:52:25.495Z" if you don't call .strict()

[colinhacks]

Makes sense - thanks for your thoughtful reply. I understand your reasoning. Closing.

====================================

P.S. @mauricedb

Seems fairly easy to implement an yup.array().nonempty() method like this that makes that logic more explicit any provides a more accurate typing information.

export interface ArraySchema<T> extends BasicArraySchema<T[]> {
    //  ...
    nonempty(): NonEmptyArraySchema<T>
}

export interface NonEmptyArraySchema<T> extends Schema<[T, ...T[]]> {
  ...
}

[colinhacks]

For folks who stumble on this:

I ended up rolling my own validation library that's more rigorous when it comes to TS type inference: https://github.com/colinhacks/zod