I am not setting up the testing execution correctly somehow, but this seems rather promising.

Codecov Report

Merging #1194 into master will increase coverage by 0.41%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #1194      +/-   ##
==========================================
+ Coverage   70.55%   70.96%   +0.41%     
==========================================
  Files         534      534              
  Lines       81675    81679       +4     
==========================================
+ Hits        57623    57965     +342     
+ Misses      24052    23714     -338

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 0f5300a...bbf1b4d. Read the comment docs.

@qtothec This looks interesting. Is this WIP right now?

@carldlaird Yes, but I'll probably need some help getting it operational. Plus we ought to have a discussion around how it would fit into the overall testing infrastructure.

It looks like this is revealing an issue in some of the trust region tests, they seem to implicitly require Ipopt.

@mrmundt @allevin @qtothec it passed!

@blnicho @qtothec @allevin - I KNOW! I might have been impatiently waiting for your PR to be merged. Thank you!

It was surprisingly fast too, 2.5 minutes to run 6962 tests (1581 skipped).

Next steps are to try it on other machines (Mac & WIn), then bump up to matrix builds across python versions.

Ok, I'm really impressed now. Only 3.5 minutes to complete "slim" test suites on Python 2.7, 3.5, 3.6, and 3.7 on Linux

Impressively quick. It's amazing. I'm about to start a run for MacOS with just Python 2.7. Let's see if the trend of success continues!

This is separate from getting this to work, but is there a way to manage the console output so that you don't have to scroll past pages of successful tests to get to the errors?

There is a lower verbosity level for nosetests but I think in most cases we would want to be able to see exactly which tests were run instead of ". . . . . . S . . . F"

I think we can take the -v off the test run to reduce the output to just the errors...

@jsiirola has pointed us to the fact that there is an old editor on macOS's named "pico" - which is triggering the pico tests to start, and thus, all fail. We'll need to address this somehow.

Replacing this with separate PRs for different distributions.