Re: [boost] [Review][PQS] Review

From: David Greene (greened_at_[hidden])
Date: 2006-06-09 13:25:58

Andy Little wrote:

> I am aware that Walter Browns original SI units could be used with various unit
> systems and units could even be converted between them, but in PQS I have
> (rightly or wrongly ) opted to only cover SI units to limit the scope of the
> project to something manageable. I realise that this is disappointing  however
> there are enough problems within one
> system ( such as the problem of linear algebra on quantities discussed in
> another thread), that I would prefer to solve them satisfactorily for one system
> rather than try to create a half-finished solution to cover every
> system.

There's a big difference between providing flexibility in the design and
actually using that flexibility to implement everything.  I would be
perfectly happy with a library that provided flexibility for other
units (decoupling prefixes, allowing non-power-of-10 prefixes, etc.)
but only actually implemented SI units.  Just as t2_quantity and
t3_quantity can come later, so can other units systems.

> I should however try to ensure that the library can potentially be extended to
> other unit systems as far as possible, and currently it is weak in that area.

I would like that very much.  I'm willing to help in whatever way
possible.  Would it help to gather a list of flexibility requirements?
The Boost Wiki would be a good place for that.

> OK. I am ware that the whole rigmarole involved in defining types and units
> could be improved. Composition with Boost Typeof would help as well as improved
> Concept docs perhaps.

I actually didn't find this too intimidating, but then I've done a fair
amount of TMP.  Of course, if it can be improved, all the better!

                             -Dave