A Brooks

I did some more tests, and parallelization's running fine! :tada: Tests consistently show parallelization is faster (although sometimes the confidence interval overlaps with it being slower), if only a little...

I noticed some cases where parallelization does indeed add overhead (I think tests ran 5 to 10 percent slower?), so I spent some time investigating. Last week, I did some...

@AndreaCrotti That's a clever solution. What changes were you thinking to make? Adding the `circleci-parallel` namespace code to Kaocha itself? While I am interested in that solution, I plan to...

@AndreaCrotti Interesting. I could see 2 being done with a Babashka script that we ship, and I think 3 could be a plugin, in theory. They'd be stretching the idea...

Threw together a simple graph of the Regal results: 

> > > I'm interested in trying this in our CI. Is a snapshot published for this branch? > > > > > > You can just point to a...

> It seems like the use of `with-redefs` in `kaocha.plugin.capture-output/with-capture` does not fit tests running in parallel... Also, `(run! f @active-buffers)` in `kaocha.plugin.capture-output/doto-capture-buffer` leads to some confusing output when tests...

I spoke with Arne about this, and given that it's been two years, we're probably not going to support classic expectations.

Along similar lines, [neil](https://github.com/babashka/neil) adds an alias to `deps.edn`.

Thank you for your thorough proposal! Two more alternatives to consider: 4. Memoize the expensive operation. I think this is as clean of a path as your alternative 3. 5....