I will take a look at it again tonight. What micro benchmark are you using to get an idea of how fast the implementations are.
Thanks Chris Damian Steer wrote: > > cdwillie76 wrote: >> I am looking to take a stab at this issue. I noticed in JIRA, JScience >> was >> mentioned as an alternative. Would that mean including the jar or just >> referencing the code for JScience. I am in the process of getting the >> specs >> to run on my mac. >> >> Thanks >> Chris > > As I understand it the reference to JScience is as a wholesale upgrading > of jruby's current numeric layer. Quite a big job, but then JScience > does the work for us. > > In this particular case you might want to try the less daunting route of > converting a BigInteger to double directly. I tried this, but the > current hack was still at least an order of magnitude faster than the > jdk version (!), so bit twiddling became less appealing. Javalution has > a toDoublePow2(long m, int n) function that would do most of the work, I > see. > > btw I've added a quick fix to the issue.[1] I suggest adding further > comments there, and visit IRC if you want to chat. > > Damian > > [1] <http://jira.codehaus.org/browse/JRUBY-3331> > > --------------------------------------------------------------------- > To unsubscribe from this list, please visit: > > http://xircles.codehaus.org/manage_email > > > > -- View this message in context: http://www.nabble.com/-jira--Created%3A-%28JRUBY-3331%29-to_f-near-Float%3A%3AMAX-tp21680921p21704305.html Sent from the JRuby - Dev mailing list archive at Nabble.com. --------------------------------------------------------------------- To unsubscribe from this list, please visit: http://xircles.codehaus.org/manage_email
