getCPUTime implementation incorrect?

Evan Laforge qdunkan at gmail.com
Tue Jun 7 02:36:10 CEST 2011


> See
>
>  http://hackage.haskell.org/trac/ghc/ticket/4247
>
> and
>
>  http://hackage.haskell.org/trac/ghc/ticket/4970
>
> We ought to fix this in 7.0.4, since getCPUTime is pretty broken on OSX
> 64-bit right now, but the fix involves an API change (to Foreign.C.Types).
>  Ian - could we apply a fix that doesn't change APIs?

At least getCPUTime could be, since its use of CTime is internal.
Adding new types to Foreign.C shouldn't break anyone, and if you
change getCPUTime to internally use the new type no one outside will
be the wiser.  Except they can remove their OS X workarounds :)



More information about the Glasgow-haskell-users mailing list