Monadic

[treeowl]

Use evaluate rather than seq to force the values to be stored. This feels right to me, but I'm not entirely confident it's the way to go.

[treeowl]

@simonmar, do you have a moment to consider whether this is the right approach?

[treeowl]

@simonmar, your attention is only requested for the second of the two commits.

[simonmar]

I have no context here, but if you're asking whether evaluate is better than seq in IO, then the answer is yes. (But I think you know that already...)

[ygale]

We decided in the discussion of PR #9 to merge this PR now as is. I am doing this by copying the branch to my repo, creating a new PR #13 , and merging that instead. That way I do not have to bother @treeowl for silly administrative things I need to do before merging, like switching CI to github actions.

Note that I am switching to the now standard practice of running CI only for the most recent few versions of GHC. This fundamental library actually supports much more than that, including versions of base stretching back into ancient history. But none of that is actually supported by GHC team anymore, It may be difficult to find platforms on which we can test those versions nowadays even if we wanted to. Being untestable, we will probably deprecate and remove support for such ancient versions someday soon.

[treeowl]

@ygale the way I see it, any GHC version not tested in CI is not actually supported. Either it should be added to CI, or the attempted support should be removed.