I just noticed that on HEAD, happy reports for Parser.y.pp: $ happy -agc --strict compiler/stage1/build/Parser.y unused rules: 2 shift/reduce conflicts: 49 reduce/reduce conflicts: 1 But the code claims: 12 October 2012 Conflicts: 43 shift/reduce 1 reduce/reduce So somewhere recently we have picked up 6 additional shift/reduce conflicts. Twan
Oh bother. I have no idea. If anyone feels able to bisect their way to the culprit, that would be good. Simon | -----Original Message----- | From: ghc-devs-bounces@haskell.org [mailto:ghc-devs-bounces@haskell.org] | On Behalf Of Twan van Laarhoven | Sent: 27 May 2013 13:00 | To: ghc-devs@haskell.org | Subject: Shift/reduce conflicts | | I just noticed that on HEAD, happy reports for Parser.y.pp: | | $ happy -agc --strict compiler/stage1/build/Parser.y | unused rules: 2 | shift/reduce conflicts: 49 | reduce/reduce conflicts: 1 | | But the code claims: | | 12 October 2012 | | Conflicts: 43 shift/reduce | 1 reduce/reduce | | So somewhere recently we have picked up 6 additional shift/reduce conflicts. | | | Twan | | _______________________________________________ | ghc-devs mailing list | ghc-devs@haskell.org | http://www.haskell.org/mailman/listinfo/ghc-devs
Hello, On Mon, May 27, 2013 at 01:01:48PM +0000, Simon Peyton-Jones wrote:
Oh bother. I have no idea. If anyone feels able to bisect their way to the culprit, that would be good.
There seems to be multiple culprits: Between http://darcs.haskell.org/ghcBuilder/builders/tn23/751 and http://darcs.haskell.org/ghcBuilder/builders/tn23/752 shift/reduce conflicts went from 43 to 45. A likely candidate is df04d2d875f4f17b04cd8bd396b62b1eadd932e8 Allow '-' to be used as an infix type constructor. Between http://darcs.haskell.org/ghcBuilder/builders/tn23/788 and http://darcs.haskell.org/ghcBuilder/builders/tn23/789 shift/reduce conflicts went to 47 with 8366792eede3c8eb486ff15d8c8e62e9363f1959 Implement overlapping type family instances. a likely candidate. Finally, between http://darcs.haskell.org/ghcBuilder/builders/tn23/806 and (there is no data for build 807) http://darcs.haskell.org/ghcBuilder/builders/tn23/808 shift/reduce conflicts went to 49 with deec5b74dee7f676b8a9f840ec7b5a813e7e0956 Be willing to parse {-# UNPACK #-} without '!' a likely candidate.
...
Best regards Thorkil
participants (3)
-
Simon Peyton-Jones -
Thorkil Naur -
Twan van Laarhoven