[Boost-users] Boost.Random: Suspicious code in uniform_smallint_integer

From: Daniel Krügler (dsp_at_[hidden])
Date: 2005-08-26 02:47:43

Although uniform_smallint_integer currently is no official code (its in
the details section and seems nowhere used), I would like to mention,
that the current implementation contains a subtle bug:

The member _factor is never initialized, thus calling operator()() would
cause U.B. The reason is, that the initializer member function set
defines a **local** variable _factor, which is never assigned to
this->_factor, such that uniform_smallint_integer::_factor will never be
assigned.

Just one question: Is detail::uniform_smallint_integer a leftover of
some experiments by the library responsible or does it exist for
interested boost users to test it?

Greetings from Bremen,

Daniel