[boost] Re: Physical Quantities revisited

From: Deane Yang (deane_yang_at_[hidden])
Date: 2003-10-24 10:20:26

Andy Little wrote:
> 
> Fish swim pretty well. That is what they are designed for.
> Impose a law requiring all fish to ride bicycles and you merely succeed in
> outlawing fish.
> 

I can't disagree with that, but from listening to the physicists explain 
their needs, I haven't heard anything that the physicists want that 
isn't also wanted by many of us non-physicists, except the desire to 
hardwire the definition of dimensions to actual physical dimensions. So 
all I am trying to express is the desire not to have the core library 
somehow tied to physical dimensions only.

> I suggest that the best road is to make a physical quantity type work in the
> field of physics.

I would agree that physics-oriented software represents the most 
important application of this library. And I do have to concede to
both you and Paul Bristow that I have nothing to show, so your 
skepticism is quite valid. I should quiet down and let the physicists 
proceed, since they definitely do have the greatest need for this 
library. And it may turn out to be what I need, too.

> A major problem is angles and rotational motion which will probably require
> a complete rethink
> 

Angles are indeed difficult to fit within the framework of a 
dimensions/units library. The basic problem is that an angle is really a 
unitless ratio. This has the consequence that there are things you want 
to be able to do with an angle that you would normally not allow for 
other physical dimensions, like feed it into a nonlinear function such 
as sine and cosine.