Nope. The problem is correctness, not speed. You can’t allow type relationships beween containers because the relationship can only be safe for either writing data or reading data, but never for both.
— John On Nov 29, 2014, at 3:55 PM, K Leo <[email protected]> wrote: > I remember somewhere it mentioned that there is a performance reason not to > have things like Array{Float64} <: Array{FloatingPoint}. Does this mean the > parametric types (defined with T) are slower? > > On 2014年11月30日 05:04, Andreas Noack wrote: >> You are welcome. It is explained better than I'm able to do here >> http://julia.readthedocs.org/en/latest/manual/methods/#parametric-methods >> >> 2014-11-29 15:59 GMT-05:00 Petr Krysl <[email protected] >> <mailto:[email protected]>>: >> >> Thanks, I hear you guys. These are the concepts, and Julia >> implements them: The two are concrete types, and hence cannot be >> derived from each other. >> >> But here it is: An array of floating-point numbers is an array of >> floating-point numbers. >> >> So how would you write the function so that it is Julia-style and >> works with any floating-point number? >> >> Thanks a lot, >> >> Petr >> >> >> >
