Re: [GHC] #4012: Compilation results are not deterministic
#4012: Compilation results are not deterministic -------------------------------------+------------------------------------- Reporter: kili | Owner: niteria Type: bug | Status: patch Priority: high | Milestone: 7.12.1 Component: Compiler | Version: 6.12.2 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Other | Unknown/Multiple Blocked By: | Test Case: Related Tickets: | Blocking: | Differential Revisions: Phab:D1073 -------------------------------------+------------------------------------- Comment (by Simon Marlow <marlowsd@…>): In [changeset:"3448f9827d2364b91bc60134aa38b74849b9d54e/ghc"]: {{{ #!CommitTicketReference repository="ghc" revision="3448f9827d2364b91bc60134aa38b74849b9d54e" Reduce non-determinism in ABI hashes with RULES and instance decls Summary: Before this change the `RULES` would be attached to one for the names from the module that appear on the left hand side. The choice depended on the `uniq` that was generated, which are known to be non-deterministic (a separate, bigger problem). Now we use `OccName`s which should be stable. Analogously for instance declarations, but they are attached to one of the types involved. Test Plan: contbuild it made `Data.Text.Internal.Fusion.Common` interface stable, previously stream fusion rule would be attached either to `streamList` or `unstreamList` depending on if the module was compiled after `cabal clean` or after `find | grep '\.o$' | xargs rm`. Reviewers: simonpj, austin, bgamari, simonmar Subscribers: puffnfresh, thomie Differential Revision: https://phabricator.haskell.org/D1073 GHC Trac Issues: #4012 }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/4012#comment:111> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler
participants (1)
-
GHC