data-reify provided the ability to turn recursive structures into explicit graphs. Many (implicitly or explicitly) recursive data structure can be given this ability, via a type class instance. This gives an alternative to using
Ref for observable sharing.
Observable sharing in general is unsafe, so we use the IO monad to bound this effect, but can be used safely even with
unsafePerformIO if some simple conditions are met. Typically this package will be used to tie the knot with DSLs that depend of observable sharing, like Lava.
Providing an instance for
MuRef is the mechanism for allowing a structure to be reified into a graph, and several examples of this are provided.
History: Version 0.1 used unsafe pointer compares. Version 0.2 of
data-reify used StableNames, and was much faster. Version 0.3 provided two versions of
MuRef, the mono-typed version, for trees of a single type, and the dynamic-typed version, for trees of different types. Version 0.4 used
Int as a synonym for
Unique rather than
Data.Unique for node ids, by popular demand. Version 0.5 merged the mono-typed and dynamic version again, by using
DynStableName, an unphantomized version of
- Fix a bug introduced in
Graphs with duplicate key-value pairs.
IntSets internally for slightly better performance.
- Fix warnings in GHC 7.10.
- Merge the mono-typed and dynamic version again, by using ‘DynStableName’, an unphantomized version of StableName.
- Use ‘Int’ as a synonym for ‘Unique’ rather than ‘Data.Unique’ for node ids, by popular demand.
- Provide two versions of ‘MuRef’, the mono-typed version, for trees of a single type, and the dynamic-typed version, for trees of different types.
- Use ‘StableName’s, making
- Use unsafe pointer compares.