2012/7/16 Andreas Mueller <[email protected]>: > On 07/16/2012 09:55 PM, Gael Varoquaux wrote: >> On Mon, Jul 16, 2012 at 01:13:55PM -0700, Jake Vanderplas wrote: >>> How do folks feel about using this sort of bleeding-edge cython syntax >>> in scikit-learn? >> I wouldn't mind waiting another year: it's not like we don't have plenty >> of things to do, and there is no urgency. >> >> That's my typical response of the guy who is advert to risk :). What do >> others think? >> > First, I'd like some benchmarks. > The functionality is marked as "experimental", which probably means > we should wait. > > Third, if they are as fast as the pointer stuff we are doing right now, > then we should probably switch all the code to the new interface. > That seems like a lot of work. As Gael said, there is already > quite a lot to do.
Still for new code I think it's better to use memoryviews rather than manual pointer manipulation that is prone to segfaulting. Also the code is cleaner and you can use syntactic sugar for many numpy-style operations on raw buffers (without the numpy python object calls overhead). -- Olivier http://twitter.com/ogrisel - http://github.com/ogrisel ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Scikit-learn-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
