[boost] Update on statistic_tests.cpp vs gcc-3.2/3.3

From: Paolo Carlini (pcarlini_at_[hidden])
Date: 2002-11-01 19:50:27

Hi Jens, hi everyone,

I have confirmed that now, thanks to your fixes, the first problem we 
were seeing has disappeared. Great!

However, with both gcc-3.2 and gcc-3.3-cvs it is still present a 
different run-time failure, which happens for *all* the tests (not only 
"minstd_rand") in "birthday spacing". For instance, for boost-cvs and 
current gcc-3.3:

paolo:~/Programming/c++/boost-cvs/boost/libs/random> a.out
Confidence level: 0.99; 1-alpha = 0.01; chi_square(19, 36.191) = 0.99
Running tests on minstd_rand
KS: 4750* [0] 9500* [0] equidistribution: 10.8 20.4   2D: 17.36 33.12
runs: up: 4750* [0] 9500* [0]   down: 4750* [0] 9500* [0]
gaps: 4750* [0] 9500* [0]
poker: 20.08 19.92
coupon collector: 15.6 13.2
permutation: 4750* [0] 9500* [0]
maximum-of-t: 15.44 12.32
birthday spacing: a.out: statistic_tests.hpp:373: void 
birthday_spacing_experiment::run(UniformRandomNumberGenerator, Counter&, 
int) const [with UniformRandomNumberGenerator = 
boost::uniform_int<boost::minstd_rand, int>, Counter = 
generic_counter<std::vector<int, std::allocator<int> > >]: Assertion 
`f.max() == static_cast<result_type>(m-1)' failed.
Abort

Can you reproduce this???

Ciao, Paolo.