[ruby-talk:03325] Re: RubyUnit assert_raiseexception

> The later ones support
> 
>   assert_exception(Name)  { code block }

While we're on it I'd like to make another request. The new version of
RubyUnit-0.2.0 doesn't include this one yet.

# 1.4142 would be enough in the first example
assert_equal_float( Math.sqrt(2), 1.41421,   4, "Approximation failed" ) 
assert_equal_float( Math.sqrt(2), 1.4142135, 7, "Approximation failed" )
assert_equal_float_rounded( Math.sqrt(2), 1.4142136, 7, "Approximation
failed" )

It would be nice to be able to say assert's for floats. The problem is that
there might be difference in calced representation and the exact value. So
it would be nice to be able to specify comparison precision (how many digits
to compare) or epsilon and the rounding method (flooring by default?).

I actually coded already the version with X digit checking, but the source
is thousand kilometers away now :(.

The other thing is to have some handling for subtests: reporting,
statistics. Last time I worked with RubyUnit the smallest unit of handling
was at test of the TestCase. It's pain in the ass if one's required to wrap
asserts into test_functions just to get proper reporting.

def test_foo            # real test
  assert_equal(1,1)     # subtest 1
  assert_equal(-1,-1)   # subtest 2
end

	- Aleksi