Re: Formal review: New Boost.Test Library

From: rogeeff (rogeeff_at_[hidden])
Date: 2001-12-06 16:18:31

> > 
> > So you mean there should be
> > 
> >     BOOST_CHECK_PREDICATE1( arg, unary_predicate)
> > and
> >     BOOST_CHECK_PREDICATE2( left, right, binary_predicate)
> > 
> > along with their relatives? Good idea. The float comparisions 
could 
> then
> > be written
> > 
> >    BOOST_CHECK_PREDICATE2(left, right, close_at_tolerance(1.0e-
5));
> >    BOOST_CHECK_PREDICATE2(left, right, weak_close_at_tolerance
(1.0e-
> 5));
> >    BOOST_CHECK_PREDICATE2(left, right, 
> >        close_at_estimated_tolerance(left, right, 
> number_of_operations));
> > etc.
> > 
> > Gennadyi, what do you think of this?
> 
> I think it's good idea. I can add this tools to the toolbax. They 
are 
> not related to floating-point comparison directly though and could 
be 
> valuable standalone. Regarding CHECK_CLOSE family: I would prefer 
to 
> keep them as short form of above statements.
> 

Oh! It's really cool. We now do not need nor BOOST_CHECK_NOT_EQUAL 
nor BOOST_CHECK_LESS nor any other binary predicate check as 
standalone tools. Also this mean that floating-point comparison could 
be easily factored out of Test Tools without loosing a functionallity 
(other than I won't be able to introduce any special printing only 
default operator<< will be used, so we will still have 1!= 1 and I 
won't show a tolerance, for there won't be such thing any more).

Do you want me to do this?

Gennadiy.