Message374537
I'm inclined to ignore this. No actual user has complained about this, and I doubt any ever will: it's got to be rare as hen's teeth to use a parameter outside of, say, [0.1, 10.0], in real life. The division error can't happen for those. For "small" alpha, the distribution simply does contain values too large to represent as binary doubles. So the only other defensible thing to do be done in this case is to return math.inf (or raise OverflowError) - but why slow it down for something no actual user cares about? |
|
Date |
User |
Action |
Args |
2020-07-28 23:35:48 | tim.peters | set | recipients:
+ tim.peters, rhettinger, mark.dickinson, serhiy.storchaka, David MacIver |
2020-07-28 23:35:48 | tim.peters | set | messageid: <1595979348.64.0.3530282796.issue41421@roundup.psfhosted.org> |
2020-07-28 23:35:48 | tim.peters | link | issue41421 messages |
2020-07-28 23:35:48 | tim.peters | create | |
|