#600 Code action to ignore hlint hints module wide

[eddiemundo]

It will also insert a -Wno-unrecognized-pragmas options pragma only if that dynflag isn't already set.

[jneira]

Amazing work, including sensible new test cases and nice refactorings, many many thanks

[jneira]

There is some problem with the pragmas placement in ubuntu and ghc-8.8.4:

 FAIL (1.88s)
      Test output was different from 'test/testdata/UnrecognizedPragmasOn.expected.hs'. Output of ["git","-c","core.fileMode=false","diff","--no-index","--text","--exit-code","test/testdata/UnrecognizedPragmasOn.expected.hs","/tmp/UnrecognizedPragmasOn.expected34086-3.actual"]:
      diff --git a/test/testdata/UnrecognizedPragmasOn.expected.hs b/tmp/UnrecognizedPragmasOn.expected34086-3.actual
      index 564503c..e65d316 100644
      --- a/test/testdata/UnrecognizedPragmasOn.expected.hs
      +++ b/tmp/UnrecognizedPragmasOn.expected34086-3.actual
      @@ -1,5 +1,5 @@
      -{-# OPTIONS_GHC -Wunrecognised-pragmas #-}
       {-# OPTIONS_GHC -Wno-unrecognised-pragmas #-}
       {-# HLINT ignore "Eta reduce" #-}
      +{-# OPTIONS_GHC -Wunrecognised-pragmas #-}
       module UnrecognizedPragmasOn where
       foo x = id x

[eddiemundo]

I think I've fixed the tests, but it's weird.

The issue was me not using the right dynflags in some cpp for the pragmas parser initialization, but that should (and does) fail the Pragmas plugin tests for 8.8.4. Yet I never saw any 8.8.4 test failures from the CI in the PR where I extracted the pragma parsing stuff into its own module. This means that the CI didn't run the Pragmas plugin tests for that PR?

Either that or somehow default dynflags have changed since then, and the Pragma plugin tests were passing back then but not now.

[jneira]

Really a great work, happy to see a contibution to my beloved hlint plugin
looking forward to your next contributions