sentenai / reinforce

Reinforcement learning in haskell
https://sentenai.github.io/reinforce/
BSD 3-Clause "New" or "Revised" License
44 stars 17 forks source link

Fix windows build and bump resolver version to the latest LTS-13.28 (ghc 8.6.5) #26

Closed arvindd closed 4 years ago

arvindd commented 4 years ago

Attention: @stites

Description

Windows build The stack.yaml file contained linux library paths for extra-libs. This was causing build errors in Windows. Hence, commented out the lines.

Bumped stack resolver version to LTS-13.28 (ghc 8.6.5) To bring it to the latest (as of today) stable stack resolver.

By bringing this to the latest version, the build failed due to a non-exhaustive-pattern match (because ghc 8.6.5 checks more exhaustive).

This error was also fixed so that the package builds with the latest resolver.

Motivation and Context

See above.

How Has This Been Tested?

Built the same both on Windows and Linux.

Screenshots (if appropriate):

Types of changes

Checklist:

stites commented 4 years ago

Pending CI, this looks fine to me -- @xnomagichash, what is the status of this repo? Is someone at Sentenai going to have time to take over maintenance?

arvindd commented 4 years ago

@stites: The CI has failed, unfortunately :-( But looks like this is due to a bad .travis.yml file: and that possibly needs another PR. Travis builds with this line: stack... reinforce-prelude, and this package is not in the current code-base (it was not there even before this PR, so obviously the CI had failed even before this PR).

I shall also fix that problem in the travis file once this PR is merged - so that will be my next PR.

stites commented 4 years ago

I shall also fix that problem in the travis file once this PR is merged - so that will be my next PR.

I just removed the offending line directly and re-triggered CI.

stites commented 4 years ago

Okay, I've built this on my machine and it is working as expected. Retriggering the build does not actually update where it is merging into master, so travis.yml remains unchanged. I would be happy to merge this but it feels a bit odd to merge without someone at @Sentenai for the final go-ahead.

The updated .travis.yml also fails on master due to some hlint suggestions (see https://travis-ci.org/Sentenai/reinforce/builds/561082616#L5880 ) in retrospect weeder and hlint in CI is a bit overkill for this project. @arvindd I think a good followup PR would make the call on removing or appeasing these.

arvindd commented 4 years ago

Ok @stites, will raise another followup PR to solve the Travis errors after this PR is merged.

arvindd commented 4 years ago

@xnomagichash: I decided to go raise this PR as I noticed that the repo had been stuck with a very old ghc / resolver, and I wanted to further this development.

Kindly merge this PR, and also drop a line on the roadmap/possibility of a handover of maintenance to somebody (including me). I can try to at least keep this afloat by accepting PRs and also furthering development / creating roadmap, of course, with the help of @stites and others who have been overseeing this development before.

If you choose to handover to me, I shall continue the maintenance in arvindd/reinforce/master and not in this repo, or even possibly move this repo under the DataHaskell project, subject to acceptance of that project.

stites commented 4 years ago

Okay, I think @xnomagichash might be a bit busy. Since I know that things are pinned correctly, I'm going to go ahead and merge this, but it would be good to get some feedback from someone at Sentenai.

Pinging some folks in case Brendan is tied up: @xilnocas, @meangrape, @brandly, @seagreen.

xnomagichash commented 4 years ago

Sorry got a little busy here. We'll discuss about who should take over maintainership.