#2528: nub not as reliable as nubBy
-------------------------------+--------------------------------------------
Reporter: jdressel | Owner:
Type: bug | Status: reopened
Priority: normal | Milestone:
Component: libraries/base | Version: 6.8.2
Severity: minor | Resolution:
Keywords: | Testcase:
Architecture: Unknown | Os: Linux
-------------------------------+--------------------------------------------
Comment (by jdressel):
Jeremy is correct: I forgot a case in my definition of (==) that makes it
not completely symmetric for second layer commutative equivalence (oops).
The relevant demonstration for the example case that failed in the linked
codepaste is:
{{{
*Main> let a = Branch (Branch (Leaf (Just 9)) Add (Leaf (Just 9))) Add
(Leaf (Just 2))
*Main> let b = Branch (Leaf (Just 9)) Add (Branch (Leaf (Just 9)) Add
(Leaf (Just 2)))
*Main> a == b
True
*Main> b == a
False
*Main> List.nubBy (==) [a,b]
[((9+9)+2)]
*Main> List.nub [a,b]
[((9+9)+2),(9+(9+2))]
}}}
I guess the lack of symmetry in (==) causes the difference in behavior,
which makes more sense to me now.
However, Adrian is also correct, the difference in behavior between "nub"
and "nubBy (==)" is what startled me. I would have found the bug in my
code eventually (especially after unit testing).
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/2528#comment:4>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler_______________________________________________
Glasgow-haskell-bugs mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
