Implement bytes() method on Blob for reading bytes into a Uint8Array

[bakkot]

See w3c/FileAPI#198. This is like the existing arrayBuffer() method, but gives you a Uint8Array instead.

Tests at web-platform-tests/wpt#46232

See also nodejs/undici#3256 for the issue for doing the same for Request/Response.

There's also a PR adding the same method to PushMessageData, but I don't think it's relevant to node. Mentioning it here anyway just in case I'm wrong.

[RedYetiDev]

@nodejs/web-standards
(Summary) this appears to be a future standard, but none of the "big 3" have taken a stance on it yet, but W3C seems to be content with it.

[benjamingr]

@RedYetiDev blob is a web standard, this method is added to the spec, we should definitely implement it (as experimental and remove experimental status when it lands).

[KhafraDev]

I won't implement it into fetch until it becomes standardized. I don't think node should land it either until then.

[RedYetiDev]

@RedYetiDev blob is a web standard, this method is added to the spec, we should definitely implement it (as experimental and remove experimental status when it lands).

I don't think it is standard yet (bytes() atleast), I agree with @KhafraDev's stance. (But, I am not a CODEOWNER)

[benjamingr]

Implementing before the spec lands is almost always a good idea since it's literally our last realistic chance to request changes or provide feedback when we see an API is problematic.

[RedYetiDev]

That's a good point, but either way, I'm not involved in that process, undici handles this, right?

[benjamingr]

Node handles this for Blob and undici handles it for request/response

[RedYetiDev]

Thanks for the info. I'm marking this as feature-request, as it is a feature-request.

[RedYetiDev]

If implemented, this would be experimental, hence the label

[KhafraDev]

almost always a good idea

There won't be anything problematic with it 😄.

[KhafraDev]

it did land in the spec already and I've opened a PR to add it to fetch

[bakkot]

The Request/Response one landed but Blob - this issue - is technically a different spec, and that PR is still open. (It's slightly more annoying because that spec lives in W3C, unlike fetch, and I haven't gone through the process of joining the relevant W3C working group.)