nathanntg / vector-math Goto Github PK

Using distinct code paths depending on whether or not object is uniquely referenced, which allows eliminating memory copy.

Add two new vector types for float and double complex vectors.

There are Accelerate functions for this, but it is not yet implemented in the API. (Need to figure out proper semantics for it.)

Add support for a circular buffer format that allows reading vectors off the end. The ideal scenario would be that it would return a Vector object, that when released, automatically advances the tail by a specified amount (not necessarily the same length of the vector). That may not be possible and there may need to be a more explicit release, or a copy on read implementation.

Provide a class that will hold the FFT / DFT setup and allow easily applying to real or complex vectors.