coot

I am not sure if I follow the suggested refactrosiation. Note that `stepInfoNonDep` and `stepInfoRaces` hold `Step`s different from the one in ``stepInfoStep`. Each thread has many `Step`s so `Map...

For that we'd need `Map StepId Step`, but that will be a large map holding all `Step`s, which has other downsides. I am not sure if optimising for pretty printing...

It would be nice to get `cabal` changes to the `haddock-project` command before we merge it. I looked into and it seems doable, although requires a bit of thinking to...

@jorisdral done; `cabal` PR haskell/cabal#9821 was merged, so with next `cabal` release we can merge it.

Could you attach simulation trace? You'd need to use `traverse_ print (Data.List.Trace.toList trace)` to print it as `ppTrace_` will evaluate it and throw an exception before returning a string.

sure, if you cannot reproduce it with the same salt then it's fixed. Is this test using `threadDelay` or `timeout`?

Please re-open if this is still an issue.

__TLDR__: a random scheduler is a nice addition but to `IOSimPOR` rather than `IOSim`. A random scheduler would be something nice to add but let me mention that one can...

I realised that if we are just interested in reordering of two events if there are `N` threads running, the chance to revert them can be even greater than `1/N`,...

If I understand you correctly you're proposing to have the same interface (`io-classes` with some possible changes) for both `io-sim` and `dejafu`. That would be definitely something good to see.