Andy:
does your recent commit to split fix this bug?
Simon
| -----Original Message-----
| From: Andy Gill [mailto:[EMAIL PROTECTED]]
| Sent: 23 January 2000 09:24
| To: GHC Bugs; Koen Claessen; John Hughes
| Subject: Problem with the GHC RNG.
|
|
| I'm trying to use QuickCheck inside GHC (has this been tried before?)
|
| -- With ghc-4.06, on NT4, the following program...
|
| import QuickCheck
| prop_toy xs = (not (null xs)) ==> label (show xs) $ xs == xs
| where
| types = (xs::[Int])
| main = test prop_toy
|
| -- Gives this output (several runs of the test program).
|
| bash% ./main
| Arguments exhausted after 4 tests.
| 50% [-4,4,-1,-3].
| 50% [-3,0,3].
| bash% ./main
| Arguments exhausted after 8 tests.
| 25% [1].
| 25% [-4,-5,0,-5,-3].
| 25% [-2,-1,2,-3].
| 25% [-1,0].
| bash% ./main
| Arguments exhausted after 0 tests.
| bash% ./main
| Arguments exhausted after 0 tests.
| bash% ./main
| OK, passed 100 tests.
| 12% [8,9,-11,12,13,6,7,8,1,2,3,-4,-3].
| 8% [4].
| 7% [1].
| 6% [3,-3].
| 6% [-5,9,20,-23,-9,22,-11].
| 6% [-3,2,0].
| 6% [-1].
| 6% [-1,6,-6].
| 4% [14,-14,-5,1,2,-7,-6,-5,-14,-13,-12,8,9].
| 4% [13,-13,21,-23,10,14,-12,21,25,-1,-27,-23,10].
| 2% [7,-1].
| 2%
| [6,24,11,17,-18,11,-24,-6,23,-12,6,-18,0,18,-6,12,-12,5,23,-12
| ,17,-18].
| 2% [40,-13,-14,41,-12,42,-11,25,-10,26,-27,27,-26].
| 2% [2,9,9].
| 2% [2,0,-3].
| 2% [2,-8,-8,-7].
| 2% [2,-2].
| 2% [14,-33,-12,-8,14,23,-24,-2,7,29,-18,-9,13,-34,-25,-3,31,22].
| 2% [11,9,-3,-2,-4,-13,12,10,1,-1,-3].
| 2% [1,3].
| 2% [1,2,3].
| 2%
| [1,-9,-28,-16,-26,-31,-41,32,27,17,7,2,-8,-18,-23,-33,26,-40,3
| 3,23,18,8,-2,-7
| ,-17].
| 2%
| [1,-5,-11,5,-1,-15,-21,22,8,2,-4,-18,-24,19,5,-1,1,9,3,-3,-17,-23,20].
| 2% [-6,-3,-6,-14,-11,13,-15].
| 2% [-6,-3,-3,8,-6].
| 2% [-11,9,-13,21,-2].
| 2% [-1,6,-6,3].
| 1% [-2,7,7,-7,0].
| bash%
|
| The final example is great; lots of good test data.
| However the tests before point to some serious shortcoming of
| the GHC RNG.
|
| Any ideas how to attack this problem?
|
| Andy Gill
|
|
