From: Eric Lemings (lemings_at_[hidden])
Date: 2006-08-21 18:47:25
> -----Original Message-----
> From: boost-bounces_at_[hidden]
> [mailto:boost-bounces_at_[hidden]] On Behalf Of Deane Yang
> Sent: Monday, August 21, 2006 4:28 PM
> To: boost_at_[hidden]
> Subject: Re: [boost] Boost Units library preview
> My main objection to this is that I am pretty sure that a
> quantity/units library that does not do implicit unit
> conversion is a *lot* easier to implement than one that does.
> And it's hard for me to believe that a library that does
> automatic unit conversion doesn't have inside of it a simpler
> non-automatic unit conversion library. So implementing
> explicit unit conversion on top of an implicit conversion
> library seems, well, backwards.
Not really. The implementation that I used is that, internally,
all arithmetic values are stored in a standard unit (typically
the SI unit). Only when an arithmetic value is assigned to or
retrieved from the object is a conversion operation actually
required. All other arithmetic operations can therefore be
oblivious to conversion operations. So doing the conversions
up front actually simplifies the implementation.