[julia-users] Re: norm() is slow when variables are arguments

[Steven G. Johnson]

On a separate note, norm([u; v]) for scalar u and v is going to be a lot 
slower than hypot(u, v) or sqrt(u*u + v*v) (although the latter does not 
check for overflow the way hypot does).   Allocating little arrays and 
calling "vector" functions for everything, even small computations on a few 
scalars, is a bad habit from Matlab.