A Brooks

A few notes: —Somewhat unusually, this is currently rebased off of Arne's cleanup branch (#233) rather than main because I thought it might benefit from some of the cleanup work...

I did a little bit more testing and realized what I have doesn't work consistently. There seems to be a nondeterministic bug with loading certain specs or vars. I don't...

Another thing to look for is places we're touching atoms or dynamic vars in code that's called within the futures. Modifying a var or atom in a secondary thread isn't...

Now it seems to be consistently giving me an error about not find the spec for `:kaocha.type/var`.

I added some retry code. I think what's happening is that there's a race condition when running `require`. ~~I suppose we could create a thread-safe version of `require`~~ Turns out...

Did a slightly more realistic benchmark with 1000 simple io tests (simply `slurp`ing `README.md`). with `:parallel`: ``` Executed in 9.39 secs fish external usr time 23.17 secs 0.00 micros 23.17...

If you have a test suite that would benefit from parallelization, I'd be interested to hear from you. **What tests do you have?** When it's a little more polished, **would...

@nwjsmith You're welcome! I somehow blanked on `test.check` as a use case even though I'm a big fan of property-based testing and we use it on some of our projects,...

Two updates: #### You can now use the `:parallel-children-exclude` key to ensure that any children will not be further parallelized. #### Setting a type of test whose children shouldn't be...

The previous test had high variance so I quit some resource-intensive programs and set the tests to use a fixed seed And with variance addresed, we get a 10 percent...