On Mon, Mar 11, 2024 at 08:43:26AM +0000, Simon Peyton Jones wrote:
* in 2.4 we could have kept the type signature of `dataD` as (it would make a plain type binder), and added a new function `dataDTyVarBnd` for the new behavior
Iavor, what about *pattern matching *on DataD? Or don't you do that?
If we exclude pattern matching on TH syntax, it becomes much more feasible to make a stable API. Is that a goal worth seeking? Would it help a few people? A lot of people?
Can't speak for anyone else, but I would find stable API combinators quite helpful. Pattern matching the constructors falls outside my modest needs. Can't say Whether that materially changes barriers to adopting new GHC releases across the ecosystem, since this would have to also simplify migration for various widely used libraries, and I hope their maintainers will contribute to the thread. -- Viktor.