On Fri, Apr 4, 2014 at 3:53 PM, Daniel Fischer <daniel.is.fischer@googlemail.com> wrote:

With -O2, all GHC versions I tried (6.12.3, 7.0.2, 7.2.2, 7.4.2, 7.6.1, 7.6.3)
produced (almost¹) the more efficient version from the USE_REPORT_PRELUDE
source, but with only -O, none did.

Also worth noting is that ghci does not not have an optimizer; this might well matter with large lists.

brandon s allbery kf8nh sine nomine associates

allbery.b@gmail.com ballbery@sinenomine.net

unix, openafs, kerberos, infrastructure, xmonad http://sinenomine.net