| Version 7 (modified by simonpj, 3 years ago) |
|---|
The new Generic Deriving mechanism (ongoing work)
GHC includes a new (in 2010) mechanism to let you write generic functions. It is described in A generic deriving mechanism for Haskell, by Magalhães, Dijkstra, Jeuring and Löh. This page sketches the specifics of the implementation; we assume you have read the paper.
This mechanism replaces the previous generic classes implementation.
Main components
- TcDeriv.tcDeriving generates an InstInfo for each data type that fulfills the isRep0 predicate. This InstInfo is the Representable0 instance for that type, allowing it to be handled generically (by kind-* generic functions).
- The representation types and core functionality of the library live on GHC.Generics (on the ghc-prim package).
- Many names have been added as known in prelude/PrelNames
- Most of the code generation is handled by types/Generics
To do
- Generate meta-information empty datatypes and instances (Datatype, Constructor, and Selector instances)
- Generate Representable1 instances
- Generic instances
- Add deriving as a keyword. This replaces the DERIVABLE pragma from the UHC implementation, and is attached to a default method on a class declaration.
- Change the Class definition to allow for generic defaults (in addition to standard defaults).
- Generate default instances for representable types which derive generic classes.
Problems/questions
- When representations are generated for more than one datatype, assembler errors appear: symbol `ghczmprim_GHCziGenerics_Representable0_closure' is already defined
- For meta-information, we need to generate a bunch of empty datatypes per user datatype declaration. Where can/should we do this? TcDeriv seems to be "too late", since we can only add instances or top-level value bindings (as the old generic mechanism did)...
- Currently, in TcDeriv.genGenericRepBind we generate instances using mkLocalInstance. Is this right, or should we use mkImportedInstance instead? SLPJ: mkLocalInstance: it's as if the instance declaration was in this module, right?
