Improve test coverage for real_problems module

[SigmaX]

We need better unit testing for our shiny benchmarks. I punted on this since the visualization examples in the docs look good, but we still need a test harness.

[SigmaX]

I've hit the low-hanging fruit, bumping coverage of this module up to 70%.

Could still use some love to explicitly test several of the evaluate() blocks for the correct output.

We do see qualitatively correct visualizations for these functions in the real-valued problems API docs, though (where matplotlib plots are generated), so the codebase has credibility. But we want that credibility to be reflected in test coverage.

[markcoletti]

Thanks for doing this, Siggy. Do you think jumping from 50 to 70% coverage is good enough for the 0.5 release?

-- Mark

On Sat, Jan 9, 2021 at 11:18 AM SigmaX notifications@github.com wrote:

I've hit the low-hanging fruit, bumping coverage of this module up to 70%.

Could still use some love to explicitly test several of the evaluate() blocks for the correct output.

We do see qualitatively correct visualizations for these functions in the real-valued problems API docs https://leap-gmu.readthedocs.io/en/latest/problem.html#real-value-problems-api, though (where matplotlib plots are generated), so the codebase has credibility. But we want that credibility to be reflected in test coverage.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/AureumChaos/LEAP/issues/32#issuecomment-757329351, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABZHWSAN6NVGZBTDURPPZ4LSZB6VPANCNFSM4KC5WHLQ .

-- mcoletti@gmail.com

[SigmaX]

Yeah, I'd say so. We should keep this issue open, and (ideally) try to get in a habit of writing a test each time we sit down to work on another feature. But I wouldn't hold up the release.