Editorial: Make better use of ValidateTypedArray #3634
Conversation
| 1. If _array_ has a [[TypedArrayName]] internal slot, then | ||
| 1. Let _taRecord_ be MakeTypedArrayWithBufferWitnessRecord(_array_, ~seq-cst~). | ||
| 1. If IsTypedArrayOutOfBounds(_taRecord_) is *true*, throw a *TypeError* exception. | ||
| 1. Let _taRecord_ be ? ValidateTypedArray(_array_, ~seq-cst~). |
There was a problem hiding this comment.
This seems worse to me. Every existing use of ValidateTypedArray is passed an object which is not yet known to be a TA.
I'd be OK with introducing a new AO for this - GetInBoundsTALengthWitness or something, maybe - and then using it ValidateTypedArray and in callsites such as this. But I really don't like re-using ValidateTypedArray for this.
There was a problem hiding this comment.
I agree with Kevin's preferred approach over the current one in the PR. As a general thing I prefer AO preconditions to never be superfluous at callsites. If there are superfluous checks, a refactoring of the kind Kevin laid out is best for clarity, so as not to raise questions for future readers.
Edit: To be clearer, the questions I'm interested in not raising for future readers are like "why is this AO called here if it doesn't need to check these conditions?" (because of the expectation that the spec strives to be as "exact" in its verbiage as it can be) and if the answer after spec archaeology is "because it was convenient" I think that's a net negative on the readability.
| 1. Else, | ||
| 1. Let _type_ be TypedArrayElementType(_typedArray_). | ||
| 1. If IsUnclampedIntegerElementType(_type_) is *false* and IsBigIntElementType(_type_) is *false*, throw a *TypeError* exception. | ||
| 1. Let _type_ be TypedArrayElementType(_typedArray_). |
There was a problem hiding this comment.
Why remove the else guard? There's no reason to do this check when _waitable_ is *true*, since it is implied by the checks in that branch.
3 participants
Split from #3629 as requested at #3629 (comment)