Ketil Malde wrote:
There was a deliberate change in strictness in 0.5 making binary strict, which apparently speeds up GHC.
Ah okay. I suppose that passes the buck to network-dns. Presumably it could be fixed fairly simply by requiring binary < 0.5 (as you suggested). Ideally, though, the implementation of its infinite random list would be changed. This would avoid having a dependency on a package which is no longer current.
Pete _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
