Re: [boost] [Probability] release of probability library with bjamsupport

From: Paul A Bristow (pbristow_at_[hidden])
Date: 2007-06-07 15:03:40

 >-----Original Message-----
>From: boost-bounces_at_[hidden] 
>[mailto:boost-bounces_at_[hidden]] On Behalf Of Brook Milligan
>Sent: 03 June 2007 19:45
>To: boost_at_[hidden]
>Subject: [boost] [Probability] release of probability library 
>with bjamsupport
>
>After figuring out the rudiments of bjam, I have released a version of
>the Probability library in the Boost Vault:
>
>    boost.probability-0.2.1.tar.gz
>
>My hope is that this makes it much easier to integrate into an
>existing Boost system for testing and use.  I would greatly appreciate
>any comments and helpful suggestions for improvement.  Once again, the
>main documentation is available at
>
>     http://biology.nmsu.edu/software/probability/
>
>Thank you very much for any helpful reviews.

As usual, bjam produced a myriad of error messages.

For MSVC 8.0

1  I found I needed to suppress

// Need to suppress warning in all examples until boost/concept_check.hpp does it (ticket #813).

#ifdef _MSC_VER
#  pragma warning(disable: 4100) // 'b' : unreferenced formal parameter
#endif

2    return probability(exp(-lambda) * pow(lambda, static_cast<double>(i)) / factorial(i));

static_cast to double or int - MSVC does not specialise for unsigned int (and say that one should not). int probably best?

3  Tests don't work with 1_34_0 version of tests.

#define BOOST_TEST_MODULE probability_tests

in place of #define BOOST_AUTO_TEST_MAIN

and #define BOOST_TEST_DYN_LINK

and BOOST_AUTO_UNIT_TEST to BOOST_AUTO_TEST_CASE (test_validator_concept) ... throughout.

4  log1p is not provided by MSVC math.h, but I added

#include <boost/math/special_functions/log1p.hpp>
  using boost::math::log1p;

5  I had to replace log(v) and exp(v) with casts like

exp(static_cast<Value>(v) to avoid ambiguity.

Autorun "j:\cpp\probability\prob\debug\probability_test.exe"
Running 24 test cases...
*** No errors detected 

for debug and release

:-))

But I also don't see why

// XXX - why can't operator<<(std::ostream&, ...) be used?

You have disabled this with
//BOOST_TEST_DONT_PRINT_LOG_VALUE (likelihood);

likelihoods.cpp
I:\boost_1_34_0\boost/test/test_tools.hpp(298) : error C2679: binary '<<' : no operator found which takes a right-hand operand of
type 'const likelihood' (or there is no acceptable conversion)
        C:\Program Files\Microsoft Visual Studio 8\VC\include\ostream(650): could be 'std::basic_ostream<_Elem,_Traits>
&std::operator <<<char,std::char_traits<char>>(std::basic_ostream<_Elem,_Traits> &,const char *)' [found using argument-dependent
lookup]
        with
        [
            _Elem=char,
            _Traits=std::char_traits<char>
        ]

...

Example 2 produces many warnings

'initializing' : conversion from 'const double' to 'float', possible loss of data

that I presume are expected as value is of type float.

Example 6 produces warning C4101: 'c' : unreferenced local variable

that could be commented out 

example 5 I'm a bit puzzled what it does/shows - none of the functions are called?

Examples 6 lots of errors like

..\..\libs\probability\test\example6.cpp(139) : error C2039: 'universal_time' : is not a member of 'boost::posix_time'
..\..\libs\probability\test\example6.cpp(139) : error C3861: 'universal_time': identifier not found
..\..\libs\probability\test\example6.cpp(143) : error C3083: 'microsec_clock': the symbol to the left of a '::' must be a type

Is this a change in 1_34_0?  I will investigate but suggestions?

I am also re-reading the documentation ;-)

And

Finally I note the thorough - indeed exemplary - use of testing and concepts :-)

Paul

---
Paul A Bristow
Prizet Farmhouse, Kendal, Cumbria UK LA8 8AB
[phone] & SMS, Mobile [phone] & SMS
pbristow_at_[hidden]