Tobias Tebbi

The confidence intervals we compute for the individual tests is actually the relative [standard error of the mean](https://en.wikipedia.org/wiki/Standard_error) of the individual runtimes. This would make perfect sense if the individual...

One way to fix this is to separate warm-up and only measure the steady-state of the system. However, this assumes that the steady-state of the VM is always the same,...

After investigating more, it turned out that the confidence interval numbers we produced were pretty much useless (details and reasoning [here](https://docs.google.com/document/d/1WCCw5WtIKXdVAz_2iBQWOPngYYVzhPM1eKDsAbtXnVI/edit)). So we disabled the feature for now.

> Wasm is independent from JS, so it makes no sense to introduce a JS type into it. JavaScript numbers are just float64 values encoded in anyref, this could be...

Thank you for the interesting points! I agree that for normal virtual dispatch, vtables are probably faster, and embedding them somehow into the RTT would also save the header word...

True, for virtual dispatch, hashing is clearly less than optimal. With a slight change in specification, one could allow the user to specify a key when creating an RTT dictionary,...

> For example, despite significant effort, no research team has been able to extend the current MVP's type system to guarantee that a String[] in Java contains only String values,...

I agree that more flexibility for initialisers would address most issues (except for the fact that there is no catch-all type for RTTs, which would be necessary to pass an...

H, I wrote a (tiny, hacky) patch for PEG.js that supports proper backtracking, as I explained here: https://github.com/pegjs/pegjs/issues/45

I just wrote a fork that supplies an environment, accessible using the variable `env`: https://github.com/tebbi/pegjs This is the same as the `#STATE{}` object suggested above. It is a quick hack,...