Closed phadej closed 7 years ago
Now that we've got rwhnf
, perhaps we should use it to cut out some boilerplate in the many NFData
instance definitions in Control.DeepSeq
? There's a whole bunch of instances that are of the form rnf !_ = ()
which we could replace with rnf = rwhnf
.
will do if/when libraries@ discussion says this is ok to go in.
For reference, the libraries mailing list discussion is here: https://mail.haskell.org/pipermail/libraries/2016-October/027385.html
Three weeks have passed since the mailing list discussion opened, and there were no objections, so I say go ahead and make the changes in https://github.com/haskell/deepseq/pull/22#issuecomment-256110833. I can merge after that.
@RyanGlScott updated this.
Resolve #3
IMHO it's just easier to have it, even recent enough generics can get it right. Some people like being implicit (it would be cool if the assumption could be checked though).