isUtf8(buffer, start?, end?)

[micnic]

Would it make sense to add 2 more arguments to isUtf8()?

declare function isUtf8(buffer: Uint8Array, start?: number, end?: number): boolean;

Where start would default to 0 and end would default to buffer.length. It should validate only the provided range, or full buffer if no arguments are provided. This would avoid function users to create a buffer slice that comes at a cost.

[marco-ippolito]

This sounds like a reasonable feature request, I'd be +1

[aduh95]

Is there an advantage for doing that rather than isUtf8(buffer.subarray(start, end))?

[micnic]

Sure, it is much faster to read a buffer in a range than to create a new buffer instance with offset from the original one. I created a benchmark here for an ASCII validation (this approach can be used for NodeBuffer.isAscii() too):

https://esbench.com/bench/65a399847ff73700a4debf72

[targos]

Your benchmark is testing an extremely small buffer. I expect the speed difference to depend largely on the buffer length (and range check length too).

[micnic]

I updated the benchmark with a larger buffer, and received similar results. Not sure we'll get any different outcome if we increase the buffer more.

https://esbench.com/bench/65a399847ff73700a4debf72

[aduh95]

OK interesting, as TypedArray.prototype.subarray does not copy memory (reuse the same underlying ArrayBuffer), I would have expected the speed to be about the same; I guess there's some overhead in creating a new TypedArray instance indeed.