Re: opt out of accursedUnutterablePerformIO

John, are you able to share this benchmarking code? Thanks, Tom On Mon, Mar 5, 2018 at 12:07 AM, John Wiegley wrote:

...

...

...

...

...

"DF" == David Feuer writes:

DF> For the record, the "public" version of runRW# is called DF> unsafeDupablePerformIO. It would be a good idea to find out if it can DF> be used throughout bytestring without a significant penalty. If so, DF> that would eliminate a bunch of scary stuff. If not, it would be nice DF> to understand why.

In my tests, using unsafeDupablePerformIO instead of accursedUnutterablePerformIO induces a 15% performance penalty.

-- John Wiegley GPG fingerprint = 4710 CF98 AF9B 327B B80F http://newartisans.com 60E1 46C4 BD1A 7AC1 4BA2 _______________________________________________ Libraries mailing list Libraries@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/libraries