I have removed the MPL: the code was there to use an int instead of a long for the factorial lookup table if the size of an int was large enough. Ultimately, the rare case in which this would be useful (when 8 <= sizeof(int) < sizeof(long)) is not worth code that won't compile on some compilers. The class now has an empty reset() member function for compatibility, and passes random_test.cpp's test_main function as well, and can serve as a drop-in replacement of the existing poisson_distribution.hpp file. Notably, operator<< and operator>> are compatible with the old version, streaming only the mean. I am unaware of how to officially submit this header, and where to change documentation. Thanks, Ross On Tue, Jan 6, 2009 at 5:19 AM, Paul A. Bristow <pbristow@hetp.u-net.com> wrote:
-----Original Message----- From: boost-bounces@lists.boost.org [mailto:boost-bounces@lists.boost.org] On Behalf Of Paul A. Bristow Sent: 05 January 2009 15:25 To: boost@lists.boost.org Subject: Re: [boost] [random] A much faster Poisson variate generation
I have also just realized that your new version using MPL may have implications for 'obselete' compilers.
It would be best if the old simpler version still worked for these compilers.
Perhaps you have already given this some thought - I see some tests for MSVC <=1300 ?
At least, it will require altering the Boost 'expected to fail' list.
Paul
--- Paul A. Bristow Prizet Farmhouse Kendal, UK LA8 8AB +44 1539 561830, mobile +44 7714330204 pbristow@hetp.u-net.com
_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost