"Phil Richards" <news@derived-software.ltd.uk> wrote in message news:20050201182448.45525F53B@derisoft.derived-software.demon.co.uk...
On 2005-01-31, Andy Little <andy@servocomm.freeserve.co.uk> wrote:
1)Unfortunately not every one has the luxury. Many incoherent units exist in many real world situations, where errors cost money and worse.
Indeed. But, again, many people only need/want dimensional analysis. My/our project is one of them. We've only been doing it with a dimensional analysis setup for 3 years, so we've probably not found all the limits yet.
2)Even within SI a quantity can be expressed in various units.
But even then there is only one dimensionality. If all you are interested in is dimensionality, then everything is hunky dory.
But you never are(OTOH maybe you are changing the impl from debug (quantity) to release((double) ) You are always interested in the value and the value must have units ( whether explicit and local or, remote and implied). I agree however that you might use an external convention and maybe this is an alternative to the very explicit units attached to a type used in the pqs UI. I guess you could select some particular unit for each of your your quantities making a set, and then apply some compiletime or possibly runtime scaling scaling factors to convert to somebody elses set. ( OTOH AFAICS the logical outcome is that you may end up with the system used in pqs.)
Really, it is that simple for many people. I accept that there are also probably many who want unit conversion and checking as well, but even they would probably like *something* rather than nothing.
Well if you dont want the range of units , simply select only the base units you want: namespace my{ typedef pqs::force::N force; typedef pqs::mass::kg mass; typedef pqs::acceleration::m_div_s2 acceleration; } using my::force; using my::mass; using my::acceleration; int main() { force f = mass(1) * acceleration(1); } regards Andy Little