| ..ah, I see, so we'll need some CPP to retain compatiblity with stable | GHCs (as parallel and deepseq have -- before your commit -- been | compatible with all stable GHC 7.x releases)... yes, I suppose so. Or, I suppose, we could revert my changes to parallel and deepseq, and add -fno-warn-redundant-constraints at the top (I suppose *that* would need cpp). And then remember in n years time to take it out. Because of the n years time issue I'm inclined to the former solution, because at least it's clear: "if compiling with GHC <= 7.10, use this signature, else that one". Simon | | | | > | > Simon | > | > | -----Original Message----- | > | From: Herbert Valerio Riedel [mailto:hvriedel@gmail.com] | > | Sent: 06 January 2015 22:03 | > | To: Simon Peyton Jones | > | Subject: breakages due to redundant constraint removals? | > | | > | Hello Simon, | > | | > | I just noticed that your recent commit to deepseq, had a devastating | > | effect: | > | | > | https://travis-ci.org/haskell/deepseq/builds/46063392 | > | | > | | > | similiarly for parallel: | > | | > | https://travis-ci.org/haskell/parallel/builds/46062982 | > | | > | | > | ...did you notice that as well? | > | | > | Cheers, | > | hvr | | -- | "Elegance is not optional" -- Richard O'Keefe