Support for typed array, DataView and ArrayBuffer in isEquals: take 2

[maartenbreddels]

supersedes #2693 fixes #2692 , feedback welcome,

[jashkenas]

Looking good! Let's add one more check to make sure ArrayBuffer is an object that exists in the runtime before starting, and I think we're there.

[maartenbreddels]

Yes I see the error. What's the best way to check. window.ArrayBuffer or root.ArrayBuffer? And I also have to skip the tests, there I don't have a reference to root, can I use window?

[jashkenas]

Why don't you test and cache a typeof ArrayBuffer check up at the top of Underscore, and then just check the variable within the deepEquals. You can't use window.

[jashkenas]

Great. Do you mind cleaning up the formatting and making it nice -- like the rest of the surrounding code? Spaces near parens, semicolons, and so on...

[maartenbreddels]

Yeah, installing eslint as we speak, the only issue that I don't know to tackle yet is node 0.8 which doesn't seem to support isView.

[coveralls]

Coverage increased (+0.04%) to 96.801% when pulling 81e5bf7 on maartenbreddels:typed_array_eq into 20e7c6e on jashkenas:master.

[jashkenas]

the only issue that I don't know to tackle yet is node 0.8 which doesn't seem to support isView.

It doesn't? That seems like a bit of a dealbreaker.

[coveralls]

Coverage increased (+0.05%) to 96.812% when pulling d73f20c on maartenbreddels:typed_array_eq into 20e7c6e on jashkenas:master.

[akre54]

Do we care to support node 0.8? Last release was 2014.

[jashkenas]

Do we care to support node 0.8? Last release was 2014.

Fair enough, then.

[jasongrout]

Do we care to support node 0.8? Last release was 2014.

Fair enough, then.

Moot point since d73f20c solves that case too, right?

[vidartf]

It seems like everyone was OK with this PR. Any chance of getting it merged?

[jgonggrijp]

@jashkenas Could you have another look and confirm that you still approve of this? If you do, I can take care of the merge conflicts.

I can also add isTypedArray to the exports and polish its implementation a little bit, if you like. Personally, I'd make that a condition for merging.

[jashkenas]

Sounds like a fine plan to me. It would also be great if we could drop that big list of toString fallback name checks, if they’re not necessary for modern JS environments anymore.