bin_buffer perf improvement?

[omarstreak]

Can you explain the reason for the Bufferish.from call and what it is doing?

We recently started using msgpack-lite in our project and in doing some performance profiling found that a large amount of time is spent read-format.js:bin_buffer. More investigation shows that on line 95 buf is a Uint8Array, and the result of Bufferish.from(buf) returns a Uint8array with the exact same data. What is the reason for the call to Buffer.from(buf)?

Given a Uint8Array Bufferish.from ends up calling its own slice. So what ends up happening is you get a redundant slice called and for lots of data that is expensive.

I made a quick hack and in our project just returned buf without doing Bufferish.from and everything worked fine. However, I think tried to modify the msgpack-lite library myself and that caused tests to fail. That obviously meant I was missing something.

My guess is that for a large number of cases the Bufferis.from call does not need to be called and would save a lot of processing time.

[omarstreak]

never mind. I read the documentation better and found the binarraybuffer code option. Works great now, thanks!