"Matthias Troyer" <troyer@itp.phys.ethz.ch> wrote in message news:5639A1B4-1BA8-4D1E-852F-0B07F3BCFC3A@itp.phys.ethz.ch...
On Aug 22, 2006, at 7:51 AM, Andy Little wrote:
"Dave Steffen" <dgsteffen@numerica.us> wrote in message news:17642.14441.683655.649056@yttrium.numerica.us...
Comment from the back row: Some people will want automatic conversions. Some people will want a compilation failure. Both are perfectly reasonable behaviors, and which one is appropriate is project specific.
If (either) units library fails to provide both capabilities, it will be unusable for some category of projects.
I don't know how to control this, nor what the interface should look like, but I am certain that a good units library will be able to go both ways.
What? By means of a Macro switch in the config header?
If we really need it, then better with a template parameter in the classes so that both can coexist
Actually, just experimenting with using this feature shows it to be quite useful. I only experimented with one example and tried changing one length unit to millimeters whereas previously it was meters. The ensuing game then was to try to get the code to compile without using any explicit unit conversions. ( You need to understand the semantics of Quan to understand that it is possible, using SI units, to work ,for example in picometers, with exactly the same accuracy as you get with meters, but you would need to study the docs ( and maybe code) in some detail to see why this is possible). I found I had to change velocity from m.s-1 to mm.s-1. Then forces to mN. I was finally left with one constant , air_density, originally with a value of 1.225 kg.m-3. which I had to change to 1.225e-9 Tg.m-3, at which point the code compiled OK without requiring any explicit conversions. (Each conversion representing a loss in speed and accuracy) It is clear from the above that the original choice of meters is probably best because of the very small air density value, but in another application perhaps different units would be a better choice. Even sticking with meters ( and 'user units' of millimeters), using the macro switch showed that some optimisation could be done by changing units of various functions and so on. So I reckon I'll leave this feature in the Quan toolbox :-) regards Andy Little