random changes

[Tim Peters]

[dan]
> I did try random.WichmannHill, which gives completely different data
> to either the new or the old version.

I expect you're confused about something, then, but there aren't enough
details to guess what.

Here's a run under 2.1.3:

>>> import random
>>> random.seed(123456789)
>>> print repr(random.random())
0.095213622685816457
>>>

Here under 2.2.3:

>>> import random
>>> random.seed(123456789)
>>> print repr(random.random())
0.095213622685816457
>>>

Here under 2.3:

>>> import random
>>> random = random.WichmannHill()
>>> random.seed(123456789)
>>> print repr(random.random())
0.095213622685816457
>>>


Same thing -- results should be (and are) bit-for-bit identical.  Now show
us what you do that leads to the "completely different data" conclusion.