Re: [boost] [GTL] redesign checked into sandbox

From: Bruno Lalande (bruno.lalande_at_[hidden])
Date: 2008-05-10 06:01:58

> John wrote:
>>Huh? I think 'Isotropic' may need some clarification, I thought you
>>meant just that distances in each coordinate's direction have the same
>>unit of measurement. Your comments make me wonder if there is something
>>more...?
>
> I do, of course, mean that distances in each coordinate's direction have
> the same unit of measurement, but I mean more than that.  Because the
> geometry is symmetric, user code can be refactored to be parameterized
> based upon abstract concepts such as horizontal/vertical orientation,
> positive/negative direction on the number line, up, down, left, right,
> positive X, negative Y directions etc.  In this way poor quality code
> with lots of flow control that is used to call the different named
> function for accessing x, y and z values of a data type can be
> parameterized by the runtime condition of what orientation/direction is
> dictated.  This is what I mean by compile time accessors (implied by
> generic order of the point) are mutually exclusive with istropic style,
> since it forces there to be flow control to choose between symmetric
> behaviors.

I'm not completely sure to see what you're talking about exactly, but
maybe it is what I had thought about as the fact of "mapping
coordinates". Let's say I have an algorithm that applies on 2D points
and I want to apply it on the 2nd and 3rd coordinates of my 3D point.
I have already thought about providing some compile-time mappers to
make such things possible. Are you talking about this king of
manipulation? (sorry if I misunderstood)

Bruno