
1 Nov
2015
1 Nov
'15
4:42 p.m.
I really don't think this is much of a flaw. It may need to be reinforced that this package (and most random libraries) doesn't provide _any_ guarantee if you use it like that (restrain the first output with randomR of lots of random generators initialized by very particular seeds sequence). The guarantee the paper will be talking about will be for a sequence of random numbers generated by one generator and this behave normally as shown by :
(minimum &&& maximum) . map (length . filter even . take 1000 . randomRs (1,10) . mkStdGen) $ [1..1000] (432,553)
In other words, if there's any problem it is only one of documentation (just a warning in the haddock ought to be sufficient). -- Jedaï