Prop validation for Component Class

[skiano]

Hello,

This is just a question.

I have a component that accepts a component constructor as a prop, and I noticed that while there is support for React.PropTypes.element, there is no PropType documented for checking if the prop is a React Class.

Am I missing something? or is there a reason why I should not use props this way?

[jimfb]

You can use ReactPropTypes.any, which basically disables typechecking and lets you pass in anything you want.

I'm not sure if it's public API (so take this with a grain of salt / it might be subject to change at any time without notice / be careful), but... there is nothing magical about the default proptype validators. They are just functions that validate a property (throwing an error if the property is invalid), and so you can write your own validator. To get started, you can look at the implementation of our default validators: https://github.com/facebook/react/blob/master/src/classic/types/ReactPropTypes.js

Closing this issue out, since it's not technically a bug. For future reference, Stackoverflow is the best place for questions that are not bugs.

[skiano]

I can write a custom validator that does what I want. I suppose my real question is weather or not React is discouraging me from using Component Constructors as props for some architectural reason?

[jedwards1211]

Yeah, I'm very surprised by this too. There have been numerous instances when I want to validate that a prop is something I can pass as the first argument to React.createElement.

[jimfb]

@skiano No, I don't think there is any fundamental architectural reason, but it shouldn't be a super common pattern. The more common pattern is to create the element on the outside and pass in the element, or to pass in a function that takes in a particular set of parameters and returns an element. This allows the owner/parent to have a little more control over the props that are passed to the nested elements, since the parent/owner is deciding the class of the element that is being created.

[zpao]

You can get more specific that any. The real reason is that this "type" is a bit hard to define and actually cover all the cases. See my comments in #5143 (comment)

[jedwards1211]

Oh, I forget, now that there are pure function components there really is no way to distinguish functions from components, is there?

[zpao]

No