[gecode-users] FloatVar implementation

[Filip Konvička]

> What you intend to do sounds very reasonable.
>   

I would prefer spending more time on analysis and design, but as I said 
we have limited time for this. So we'll try to do "as little as we need" 
but keep things generic so others can easily extend our work.

> As it comes to integration versus contribution: we would wish integration
> but we might have to settle for contribution. We can or want to integrate
> only if:
>  - The code is not tainted by a too restricted license, so only MIT.
>   

I got the approval, and we can use that license. :-)

>  - The code is of high quality such that it is well tested, well
>    documented, useful for many people, and realistic as it comes to
>    performance. We will not integrate code that we can not fix or port
>    ourselves.
>   

I think that the initial version should be quite small. We're used to 
making tests and generating doxygen docs, so that should be fine. We'll 
try to stick to your code style (ours is very similar).

>  - The code must be portable and conform to the C++ standard.
>   

Dtto. We need our code work at least with MSVC 2008 and gcc 4.3.0, and 
we try to stick to the C++ standard.

> I believe that you should make it for all three criteria. But let's see. We
> can first go for contribution and later promote to integration.
>   

That would be perfect. I prefer contribution for start, and when things 
get stable we can integrate, in a year or so. We'll also see how many 
people are actually interested in this, but I think at least it could be 
used for GeOz, and maybe packaged with GecodeJ.

Cheers,
Filip