- Switch from ShortText to Text. In many cases, this makes it possible for the resulting Value to share the input byte array instead of allocating additional memory.
- Get rid of word-compat. Require new GHC instead.
- Optimize decoding. Time for
json/twitter/100/decodedrops from 190us to 150us.
- Fix build with GHC 9.8.
- Bumped upper bound on
transformersto <0.7. - Bumped upper bound on
textto <2.2.
- Update package metadata.
- Add
decodeNewlineDelimited. - Add
toChunks,toBytes,toText,toShortText,toByteArray.
- Correct implementations of
object15andobject16.
- Add
objectFromListandarrayFromList.
- Add
object(13|14|15|16|17). - Add
ToValueinstances forWord,Text,Value,Scientific, list (i.e.[]), the unit type (i.e.()), - Add
textandshortTextfor value construction.
- Add typeclass
ToValuefor encoding. - Add functions
int,(int|word)(8|16|32|64),boolfor constructing objects.
- Add
Json.Flattenmodule. - Drop support for GHCs older than 9.0.
- Replace integer-gmp with ghc-bignum.
- Build with GHC 9.2.3.
- Test suite now requires aeson >= 2.0 instead of < 2.0.
- Support Jackson's SMILE format as an encode target.
- Use
bytebuild'srebuildfunction for 2x perf improvement on encode. - Bump bytebuild for buffer overflow fix.
- Switch from
ChunkstoSmallArrayin theObjectandArraydata constructors. This makes the library simpler to use but it a breaking change. - Expose
emptyArrayandemptyObject. - Add
object(9|10|11|12)as convenience helpers for construction.
- Add infix pattern synonym for
Member. - Add
object(1|2|3|4|5|6|7|8)as convenience helpers for construction.
- Add
encode.
- Initial release.