A proxy class between std random distributions and GSL RNG

[Peter]

Boost provides many classes for random number distributions and they are flexible so users can pass their random number generator. I would like to use the singleton in yat::random and to do that I need a proxy class that uses yat::random::RNG and that behaves as a ​UniformRandomNumberGenerator. yat::random::DiscreteUniform is a candidate with some extensions or we need a new class solely for this purpose.

[Peter]

ticket #869 was marked related

[Peter]

Would be nice if the proxy works against random distribution classes in ​std <random> (c++11)

[Peter]

Replying to Peter:

Would be nice if the proxy works against random distribution classes in ​std <random> (c++11)

correct url: ​http://www.cplusplus.com/reference/random/

[Peter]

In 3897:

refs #869 and #858

[Peter]

In 3898:

closes #858 and #869

[Peter]

Since we require c++11 now (ticket #949), I suggest we replace this implementation against boost with an implementation against std (before this gets released). Also, means we can get rid of the configure checks for related boost headers with certain distribution as well as merging those tests back to one test file (ticket #944)

[Peter]

In 3958:

change to using std distributions instead of boost. Some distributions in boost are not provided in std, but it's easy for a user to use boost distributions as template parameter in random::Distribution and it daughter classes. closes #858.

[Peter]

In 4008:

Remove Distribution class and its daughter classes. It used
DiscreteUniform? as an Engine with std random distributions but std
requires that an Engine has ::min() and ::max() functions declared
constexpr and the code therefore does not compiler against clang's
implementation of stdlib and it is not possible to implement the
functions constrexpr without significant changes in the RNG class.

refs #858

[Peter]

Keep this open as an idea.