Michael Mior

@jviotti Done!

As an update, I get the same result if I use debug build (same error messages and no assertions seem to be hit). I also tried building with Clang 12...

I can't think of any specific configuration that's likely to be problematic (although there's obviously something). I'll try to give this a shot on another machine sometime soon. In the...

I tried again and I still get what look like pretty similar test failures unfortunately.

Logs are below. I'm using Clang 18.0.0. ``` cmake -E env UBSAN_OPTIONS=print_stacktrace=1 \ ctest --test-dir ./build --build-config Debug \ --output-on-failure --progress --parallel Internal ctest changing into directory: /home/mmior/apps/jsonbinpack/build Test project...

@Leonolovich This is discussed above. There are a few different approaches. You would have to change the range of `b` from 2 to 7. Then you would check the condition...

@dtommandru hyperopt supports whatever algorithm you like for suggesting new runs. The built-in algorithms include TPE, random, annealing, as well as a mixture of any of the above. I would...

As a suggestion, if some of the other cleanup can be done without Python 2 support getting in the way, why not have a final Python 2 release for those...

Why not just avoid specifying `max_evals` entirely? (Serious question, but it seems like hyperopt should stop on it's own when no new values are produced.)

Check out the wiki on [using MongoDB](https://github.com/hyperopt/hyperopt/wiki/Parallelizing-Evaluations-During-Search-via-MongoDB). In addition to enabling parallelization, any trials which complete will be recorded and not repeated when the optimization is run again.