Oliver Beckstein

The majority opinion appears to be to **deprecate `results.times` and `results.frames` as produced by `AnalysisFromFunction`** and then remove for 3.0 for consistency.

@tylerjereddy can you keep an eye on this PR, please? If you don't have the bandwidth, please unassign yourself again. Thanks.

Sounds good @tylerjereddy !

Pushed out so that open PRs and forks don't get upset; I'd rather move along than add more friction.

I was able to get normal coverage when running locally ``` pytest -v --disable-pytest-warnings --durations=20 --low-performance --cov-report=xml --cov-report=term-missing --color=yes ./tests coverage report ``` I can locally reproduce the *No data...

The issue was that `--cov=gromacs` in the presence of a `./gromacs` directory expected the code to be installed with `pip install -e .` instead of a normal installation. Although the...

I’m not sure how to write the tests so that they work with different call signatures across different versions. Try/except? @tylerjereddy , any hints? You have a lot of experience...

Note that the only option for fixing this issue is to break the existing API: we need the `results` attribute to now hold a `Results` instance instead of a numpy...

My view is that this was a mistake in the first place as it contradicts our specs and we should treat it as a bug and thus fix it. This...

> I could try to see if it is possible to adjust it to get Results if that's fine, especially if there are a couple of months time to adjust...