graphviz-2999.20.0.1 does not compile with GHC 8.0.x

[peti]

Citing from https://hydra.nixos.org/build/74574307:

Data/GraphViz/Printing.hs:135:13: error:
    Ambiguous occurrence ‘<>’
    It could refer to either ‘PP.<>’,
                             imported from ‘Text.PrettyPrint.Leijen.Text.Monadic’ at Data/GraphViz/Printing.hs:(79,1)-(85,68)
                          or ‘Data.Semigroup.<>’,
                             imported from ‘Data.Semigroup’ at Data/GraphViz/Printing.hs:105:24-36

Data/GraphViz/Printing.hs:285:16: error:
    Ambiguous occurrence ‘<>’
    It could refer to either ‘PP.<>’,
                             imported from ‘Text.PrettyPrint.Leijen.Text.Monadic’ at Data/GraphViz/Printing.hs:(79,1)-(85,68)
                          or ‘Data.Semigroup.<>’,
                             imported from ‘Data.Semigroup’ at Data/GraphViz/Printing.hs:105:24-36

Data/GraphViz/Printing.hs:285:21: error:
    Ambiguous occurrence ‘<>’
    It could refer to either ‘PP.<>’,
                             imported from ‘Text.PrettyPrint.Leijen.Text.Monadic’ at Data/GraphViz/Printing.hs:(79,1)-(85,68)
                          or ‘Data.Semigroup.<>’,
                             imported from ‘Data.Semigroup’ at Data/GraphViz/Printing.hs:105:24-36

Data/GraphViz/Printing.hs:288:26: error:
    Ambiguous occurrence ‘<>’
    It could refer to either ‘PP.<>’,
                             imported from ‘Text.PrettyPrint.Leijen.Text.Monadic’ at Data/GraphViz/Printing.hs:(79,1)-(85,68)
                          or ‘Data.Semigroup.<>’,
                             imported from ‘Data.Semigroup’ at Data/GraphViz/Printing.hs:105:24-36

Data/GraphViz/Printing.hs:288:35: error:
    Ambiguous occurrence ‘<>’
    It could refer to either ‘PP.<>’,
                             imported from ‘Text.PrettyPrint.Leijen.Text.Monadic’ at Data/GraphViz/Printing.hs:(79,1)-(85,68)
                          or ‘Data.Semigroup.<>’,
                             imported from ‘Data.Semigroup’ at Data/GraphViz/Printing.hs:105:24-36

Data/GraphViz/Printing.hs:291:25: error:
    Ambiguous occurrence ‘<>’
    It could refer to either ‘PP.<>’,
                             imported from ‘Text.PrettyPrint.Leijen.Text.Monadic’ at Data/GraphViz/Printing.hs:(79,1)-(85,68)
                          or ‘Data.Semigroup.<>’,
                             imported from ‘Data.Semigroup’ at Data/GraphViz/Printing.hs:105:24-36

Data/GraphViz/Printing.hs:291:35: error:
    Ambiguous occurrence ‘<>’
    It could refer to either ‘PP.<>’,
                             imported from ‘Text.PrettyPrint.Leijen.Text.Monadic’ at Data/GraphViz/Printing.hs:(79,1)-(85,68)
                          or ‘Data.Semigroup.<>’,
                             imported from ‘Data.Semigroup’ at Data/GraphViz/Printing.hs:105:24-36

Data/GraphViz/Printing.hs:348:37: error:
    Ambiguous occurrence ‘<>’
    It could refer to either ‘PP.<>’,
                             imported from ‘Text.PrettyPrint.Leijen.Text.Monadic’ at Data/GraphViz/Printing.hs:(79,1)-(85,68)
                          or ‘Data.Semigroup.<>’,
                             imported from ‘Data.Semigroup’ at Data/GraphViz/Printing.hs:105:24-36

[ivan-m]

That's odd, because it builds for me on Travis-CI.

Is the Nix expression doing any jailbreaking, etc.?

Note that it should be using wl-pprint-text-1.2.* (as it no longer exports Text.PrettyPrint.Leijen.Text.Monadic.<>), which makes me think this is due to removal of version bounds.

[peti]

You are right, we still had a jailbreak in place in Nix to work around the constraint on previous versions of QuickCheck, but I see that issue no longer exists. When we build the package with wl-pprint-text-1.2.*, the build succeeds, indeed. I do run into test suite failures, however, because the test suite has an undeclared dependency on the neato executable:

  tests/Data/GraphVizSpec.hs:27:
  1) Data.GraphViz FGL Graphs are augmentable
       uncaught exception: IOException of type NoSuchThing (neato: runInteractiveProcess: runInteractiveProcess: exec: does not exist (No such file or directory)) (after 1 test)
       mkGraph [] []

  tests/Data/GraphVizSpec.hs:29:
  2) Data.GraphViz Ensure augmentation is valid
       uncaught exception: IOException of type NoSuchThing (neato: runInteractiveProcess: runInteractiveProcess: exec: does not exist (No such file or directory)) (after 1 test)
       mkGraph [] []

  tests/Data/GraphVizSpec.hs:31:
  3) Data.GraphViz Unique edges in augmented FGL Graphs
       uncaught exception: IOException of type NoSuchThing (neato: runInteractiveProcess: runInteractiveProcess: exec: does not exist (No such file or directory)) (after 1 test)
       mkGraph [] []

[peti]

Fixed in peti/nixpkgs@28c30b8. Thank you for your help!

[ivan-m]

Not a problem.

Regarding needing pkgs.graphviz, would it be more convenient if the test suite detected if dot and neato were available before running those tests? The main reason I haven't so far is that I wanted to make sure my Travis-CI builds failed when working out how to add them in rather than succeed without running those tests (I was thinking maybe a Cabal flag that's enabled by default adding in the option of doing checking).