Adam Rice

For reference, here's the benchmark I'm using: https://gist.github.com/ricea/ca766d90ba985dfb7e1cd54bba241565 If you want to run it, don't open the console until the benchmark is over or it will mess up the results....

I added support for backpressure for sync iterators (calling `next()` inside pullAlgorithm) to see how much difference it would make. I also redid my measurements with a million iterations. Behaviour...

Sorry for my radio silence on this. I'm leaning towards keeping the ES-standard sync-to-async behaviour. Even though we know the performance is suboptimal, it's probably acceptable in almost all cases....

One way forward would be if leaving `reset()` undefined on the transformer would just disable the reset functionality altogether. If a transformer really wanted a reset that just cleared the...

Update from 2017: hints about good chunk size propagate down a pipe via desiredSize, but "narrow" sections of the pipe or non-byte sections will lose information. There is some [language...

Assuming we want chunks to be transferred where possible, rather than copied, we have three main options on the table: ### 1. **Opportunistic greedy transfer.** Any part of the chunk...

> For (3), I'm assuming the protocol would be a function of some sort. (Either one that returns the transferable parts, or one that somehow "does the transfer".) Doesn't that...

I had a discussion with @yutakahirano about transferring strategies. He proposed the mental model of a transfer consisting of creating a special kind of TransformStream with one leg in the...

> An alternate approach would be to do two transfers: one same-realm that transfers from the producer's control into the internal queue on the sender side, and one that transfers...

> I'll be happy if postMessage doesn't care about IsDisturbed. Yes. I've thought about it some more and I don't think there's a need for it to look at IsDisturbed.