Because of its untyped internal type, it copes with type changes:
As a bonus, multiline Strings and Text are diffed linewise:
Changes
0.3.4
Remove ‘Eq a’ requirement from ‘ediffGolden’. The constraint wasn’t used, as we only compare ‘Expr’ representations.
0.3.3
Change ‘ediffGolden’ so that parse errors in expected file don’t cause the hard failure.
This way you may --accept new results even when expected files are broken, e.g. due merge conflict markers.
For now the change is a bit a hack to avoid breaking change in type-signature of ediffGolden/1.
0.3.2
Add ‘ediffGolden1’, a variant of ‘ediffGolden’ with an additional argument.
0.3.1
Support GHC-8.6.5…9.10.1
0.3
Breaking change:
Make HashSet and HashMap ToExpr instances sort the resulting
lists of expressions.
This makes the results deterministic.
… but your golden files will need adjustment.
https://github.com/haskellari/tree-diff/issues/67
Add Ord Expr and Ord OMap instances
Depend on data-array-byte to provide more ByteArray instances
0.2.2
Add instances for base and primitive’s ByteArrays.
0.2.1.1
Support base-4.16 (GHC-9.2)
0.2.1
Support aeson-2.0.0.0
0.2
Change the pretty printing to use less horizontal space.
Pretty datastructure is changed.
Change Expr to use OMap; pretty-printing preserves field order.
Add strict and these instances
Add Eq and NFData (Edit a) instances.
Bump lower bounds
0.1
Support GHC-7.4 … 8.8 (use allow-newer for GHC-8.8-alpha).
Use raw GHC.Generics (drop generics-sop dependency)
Use own memoising (Vector lookup, drop MemoTrie dependency)
Singleton data-types (both data and newtypes) are printed in App form (i.e. no-record).
Change license to GPL-2.0-or-later
0.0.2.1
Support semigroups-0.19, hashable-1.3 and generics-sop-0.5
0.0.2
Print compact diffs
Add ToExpr instance for ShortByteString
0.0.1
Use BS.readFile and BS.writeFile in ediffGolden.
This makes files read and written in UTF8 independently of locale.
Fixes builds on appveyor.
