Support es6 symbols in proptype

[dashed]

Would you be open to support es6 symbols in proptype?

At the moment I'm testing them as typeof Symbol() === 'symbol' in custom proptype functions.

[steve-taylor]

Here's what I use (not sure if it's the right approach):

function symbol(props, propName, componentName) {
    const prop = props[propName];
    if (prop && (typeof prop !== 'symbol')) {
        return new Error(`[${componentName}]: Expected property ${propName} to be a Symbol, but its actual type is ${typeof prop}.`);
    }
}

symbol.isRequired = function(props, propName, componentName) {
    return props[propName] ?
           symbol(props, propName, componentName) :
           new Error(`[${componentName}]: Property ${propName} is required.`);
};

export default symbol

[zpao]

We've generally said no to a lot of custom proptypes checks but this one feels like it's probably ok. In theory it's a 1 line change. However we aren't applying the Symbol transform (which changes all typeof checks) so we'd have to do something similar with the type (already have that in place for RegExps and old browsers) - not hard but just a little bit more work.

Thoughts @sebmarkbage @spicyj

[sebmarkbage]

Yea. Seems reasonable. The .constructor === Symbol as a fallback is ok. Not ideal since it doesn't work cross-realm but we should use whatever is standard practice among popular polyfills.

[puradox]

Any possibility that this will be added in future versions?

[zpao]

Yes, we just never got back to it. If you're interested in working on it, this is something that would make a nice addition in 15.1.

[RaitoBezarius]

@zpao No one is assigned? I could do a PR for it!

[gaearon]

GitHub doesn’t make it very obvious 😢

Usually when people start working on a feature they reference it, and it appears in the issue.
You can see the existing PR above:

[zpao]

Also, you can't assign issues to people who don't have commit access, making it much harder to communicate that somebody outside the core team is working on it.

[RaitoBezarius]

That's a shame, it seems related to isaacs/github#369 and isaacs/github#100 :/

[ffxsam]

Any update on this? I suppose I'll switch to string constants in this one instance, but symbols would've been better!

[RaitoBezarius]

@ffxsam A PR is running on #6377 -- I don't know what is the status, but it looks complete on our side. Maybe, it will be merged soon!

[gaearon]

We’re still ironing out how we’ll move on with cutting branches on 15.x but once it’s decided I expect this to get in.

[sophiebits]

Looks like #6377 was merged. Should be in 15.1 or 15.2.