speculation/CHANGELOG.markdown
2016-01-17 00:43:03 -05:00

68 lines
1.4 KiB
Markdown

/1.5.0.3/
* Support GHC 8
* Support `transformers` 0.5
/1.5.0.2/
* Build warning-free on GHC 7.10+
* Added an HLint configuration
* Removed a redundant constraint from the type of `sequenceByA_`.
* Removed a redundant `Monad m` constraint from `instance MonadSpec (ContT r m)`.
/1.5/
* Removed the use of `tag-bits`. This enables the API to be `Trustworthy`.
/1.4/
* Simplified MonadSpec
/1.3/
* Removed old benchmark/test framework.
* Make numSparks support mandatory.
* Moved to Control.Concurrent from Data
* Added MonadSpec, so we can make instances for Codensity, etc. in other packages
* Removed the ContT r STM combinators
* Cleaned out old issues
* Made compatible with the removal of the Eq/Show superclasses of Num in GHC 7.3+
/1.2.0.2/
* Fixed name collision with the new Distribution.Simple.testHook in Setup.lhs
/1.2.0.1/
* Weakened dependencies
/1.2.0.0/:
* Reorganized the module hierarchy into Data.Speculation
/1.1.0.0/:
* Added support for numSparks
/1.0.0.0/:
* Released
/0.9.0/:
* Removed interim boxing in the unsafeIsEvaluated and unsafeGetTagBits check
/0.8.1/:
* Added Data.List.Foldable
* Added Data.Traversable.Foldable
* Fixed an off-by-one error in the arguments to the speculative fold estimators
/0.8.0.2/:
* changed tests and benchmarks to not build by default to work around corruption in the hackage db
/0.8.0.1/:
* test suite now forces build