error Test.QuickCheck.resize: negative size since 0.2.1

[hasufell]

GHCup uses generic-arbitrary: https://gitlab.haskell.org/haskell/ghcup-hs/-/blob/master/test/GHCup/ArbitraryTypes.hs

Since 0.2.1 the tests fail with:

Test suite ghcup-test: RUNNING...
FF
Failures:

  src/Test/Aeson/Internal/RoundtripSpecs.hs:59:5:
  1) GHCup.Types.JSON, JSON encoding of GHCupInfo, allows to encode values with aeson and read them back
       uncaught exception: ErrorCall
       Test.QuickCheck.resize: negative size
       CallStack (from HasCallStack):
         error, called at src/Test/QuickCheck/Gen.hs:135:22 in QuickCheck-2.14.2-a4039abb5c2bf61956f8dce1084c880ba597c6b14935be0ae502da628dbe9111:Test.QuickCheck.Gen
       (after 1 test)
         Exception thrown while showing test case:
           Test.QuickCheck.resize: negative size
           CallStack (from HasCallStack):
             error, called at src/Test/QuickCheck/Gen.hs:135:22 in QuickCheck-2.14.2-a4039abb5c2bf61956f8dce1084c880ba597c6b14935be0ae502da628dbe9111:Test.QuickCheck.Gen


  To rerun use: --match "/GHCup.Types.JSON/JSON encoding of GHCupInfo/allows to encode values with aeson and read them back/"

  src/Test/QuickCheck/Gen.hs:135:22:
  2) GHCup.Types.JSON, JSON encoding of GHCupInfo, produces the same JSON as is found in test/golden/GHCupInfo.json
       uncaught exception: ErrorCall
       Test.QuickCheck.resize: negative size
       CallStack (from HasCallStack):
         error, called at src/Test/QuickCheck/Gen.hs:135:22 in QuickCheck-2.14.2-a4039abb5c2bf61956f8dce1084c880ba597c6b14935be0ae502da628dbe9111:Test.QuickCheck.Gen

  To rerun use: --match "/GHCup.Types.JSON/JSON encoding of GHCupInfo/produces the same JSON as is found in test/golden/GHCupInfo.json/"

Randomized with seed 25004185

Finished in 0.2310 seconds
2 examples, 2 failures

I'm suspecting it may be 0e2e124

[eborden]

This is blocking our (Freckle) use of generic-arbitrary in stackage LTS 19.3. Can we provide any help in getting the fix PRs (#10) merged and a new release cut?

[eborden]

Is there any movement to get this fixed? Do you need any help cutting a release with the patch?

[ilyakooo0]

@eborden Thanks for bringing this issue up again

0.2.2 is now released with the fix: https://hackage.haskell.org/package/generic-arbitrary-0.2.2