Test suite depends on unstable ordering returned from aeson

[sternenseemann]

Test failure should be reproducible with cabal v2-test --constraint 'unordered-containers >= 0.2.16, as that release changed ordering returned from HashMaps via toList, folds etc.

Running 1 test suites...
Test suite json-to-haskell-test: RUNNING...

analyze
  should build a record with all types FAILED [1]
  should share records definitions for identical subrecords FAILED [2]
  should pick good names for differing records which share field names FAILED [3]
  should pick the best name if there are multiple possible names but some conflict FAILED [4]
  should sanitize weird field and record names

Failures:

  test/Spec.hs:15:5: 
  1) analyze should build a record with all types
       expected: "data Model = Model\n  { modelA :: Double\n  , modelD :: Double\n  , modelB :: Text\n  , modelC :: [Maybe Value]\n  } deriving (Show, Eq, Ord)"
        but got: "data Model = Model\n  { modelB :: Text\n  , modelA :: Double\n  , modelD :: Double\n  , modelC :: [Maybe Value]\n  } deriving (Show, Eq, Ord)"

  To rerun use: --match "/analyze/should build a record with all types/"

  test/Spec.hs:15:5: 
  2) analyze should share records definitions for identical subrecords
       expected: "data A = A\n  { aAge :: Double\n  , aName :: Text\n  } deriving (Show, Eq, Ord)\n\ndata Model = Model\n  { modelA :: A\n  , modelB :: A\n  } deriving (Show, Eq, Ord)"
        but got: "data A = A\n  { aAge :: Double\n  , aName :: Text\n  } deriving (Show, Eq, Ord)\n\ndata Model = Model\n  { modelB :: A\n  , modelA :: A\n  } deriving (Show, Eq, Ord)"

  To rerun use: --match "/analyze/should share records definitions for identical subrecords/"

  test/Spec.hs:15:5: 
  3) analyze should pick good names for differing records which share field names
       expected: "data A = A\n  { aField :: Field\n  } deriving (Show, Eq, Ord)\n\ndata B = B\n  { bField :: Field2\n  } deriving (Show, Eq, Ord)\n\ndata Field = Field\n  { fieldName :: Text\n  } deriving (Show, Eq, Ord)\n\ndata Field2 = Field2\n  { field2Age :: Double\n  } deriving (Show, Eq, Ord)\n\ndata Model = Model\n  { modelA :: A\n  , modelB :: B\n  } deriving (Show, Eq, Ord)"
        but got: "data A = A\n  { aField :: Field\n  } deriving (Show, Eq, Ord)\n\ndata B = B\n  { bField :: Field2\n  } deriving (Show, Eq, Ord)\n\ndata Field = Field\n  { fieldName :: Text\n  } deriving (Show, Eq, Ord)\n\ndata Field2 = Field2\n  { field2Age :: Double\n  } deriving (Show, Eq, Ord)\n\ndata Model = Model\n  { modelB :: B\n  , modelA :: A\n  } deriving (Show, Eq, Ord)"

  To rerun use: --match "/analyze/should pick good names for differing records which share field names/"

  test/Spec.hs:15:5: 
  4) analyze should pick the best name if there are multiple possible names but some conflict
       expected: "data A = A\n  { aField :: Field\n  } deriving (Show, Eq, Ord)\n\ndata B = B\n  { bField :: Other\n  } deriving (Show, Eq, Ord)\n\ndata C = C\n  { cOther :: Other\n  } deriving (Show, Eq, Ord)\n\ndata Field = Field\n  { fieldName :: Text\n  } deriving (Show, Eq, Ord)\n\ndata Model = Model\n  { modelA :: A\n  , modelB :: B\n  , modelC :: C\n  } deriving (Show, Eq, Ord)\n\ndata Other = Other\n  { otherAge :: Double\n  } deriving (Show, Eq, Ord)"
        but got: "data A = A\n  { aField :: Field\n  } deriving (Show, Eq, Ord)\n\ndata B = B\n  { bField :: Other\n  } deriving (Show, Eq, Ord)\n\ndata C = C\n  { cOther :: Other\n  } deriving (Show, Eq, Ord)\n\ndata Field = Field\n  { fieldName :: Text\n  } deriving (Show, Eq, Ord)\n\ndata Model = Model\n  { modelB :: B\n  , modelA :: A\n  , modelC :: C\n  } deriving (Show, Eq, Ord)\n\ndata Other = Other\n  { otherAge :: Double\n  } deriving (Show, Eq, Ord)"

  To rerun use: --match "/analyze/should pick the best name if there are multiple possible names but some conflict/"

Randomized with seed 1870694535

Finished in 0.0341 seconds
5 examples, 4 failures
Test suite json-to-haskell-test: FAIL
Test suite logged to:
dist/test/json-to-haskell-0.1.1.2-json-to-haskell-test.log
0 of 1 test suites (0 of 1 test cases) passed.