Jeffrey Young

I'd prefer to use `lvish` as inspiration and as a reference rather than directly using it in the project. IMHO we should have a separate implementation because we have different...

> Block context acquisition is the time between these two calls to trace. This seems like a good candidate for inserting cost centers by hand then: https://downloads.haskell.org/ghc/latest/docs/users_guide/profiling.html#inserting-cost-centres-by-hand and then with...

> A separate variance analysis demonstrated that that calculation varies by no more than 1ms. This can only be true if the distribution of samples is found to be normal....

> you can verify if the approach is sound. Sure, if you can send me the data I'll throw it into R and check the normality.

Ah, I just saw that `System.GHCJS` exists. Perhaps this is just a case of patching `Setup.hs` then?

Reading through the Setup.hs more carefully I don't think this error is actually meaningful. Its thrown by `emcc` during `checkRDRAND` but the result will be caught by `checkGetRandom` which compiles...

Alright I was able to show the names of the value closures by adding some `Text` to `HO`. The problem now is that the error messages for the tests seem...

> How about using the SrcLoc of the lambda to identify it? This would work even if the lambda is never given a name. Yes I wanted to go down...

On second thought, I think that this printing `thing` is a bug. It prints the argument name because of this line: ``` printKont e (InArg fun _env k) = text...

Note that if instead of requiring the function symbol being fully evaluated before the argument like this: ``` -- functions InArg :: !(CoreF TypePattern ConstructorPattern Core) -> !VEnv -> !Kont...