#9315: Weird change in allocation numbers of T9203 -------------------------------------+------------------------------------- Reporter: nomeata | Owner: simonmar Type: bug | Status: new Priority: normal | Milestone: Component: Runtime | Version: 7.9 System | Keywords: Resolution: | Operating System: Unknown/Multiple Differential Revisions: | Type of failure: None/Unknown Architecture: | Test Case: Unknown/Multiple | Blocking: Difficulty: Unknown | Blocked By: | Related Tickets: | -------------------------------------+------------------------------------- Comment (by nomeata): Sigh. Both `Foreign.Storable` and `GHC.Fingerprint` have (with `-dsuppress-unique`) identical core. Also `Data.Typeable.Internal` where the ticky diffs show the same number of calls to `base:Data.Typeable.Internal.$fTypeableks_$ctypeRep#`, but a different allocation count.... hah! but only when I copy the ghc invocation from one host to the other, not if if I run `make` there, which for some reason passes `-O2` when compiling `Data.Typeable.Internal`.... So the problem is as follows: Invocation of `validate` and the usual `devel2` settings in `mk/build.mk` have `GhcLibHcOpts += -O -dcore- lint`, while the default in the absence of everything is `-O2` (in `in GhcLibHcOpts=-O2`). And sometime that makes a difference... Clearly, this needs to be consolidated. What are the settings used to actually build the releases? `-O` or `-O2`? The appropriate one needs to be used in `validate`, otherwise these tests are not very useful. And then people will have to use the same flags in their `mk/build.mk` or learn to ignore benchmark results from such a tree. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/9315#comment:5 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler