Hi,
I experimented with this a bit, in particular the following function
generates output:
check unused = head [(int,x,x2) | x <- [-10000..10000], let int = tn
x, let x2 = ti int, x2 /= x]
(the unused is just to stop hugs saving the CAF)
This generates some output, even though it doesn't under GHC. The
output I got first was:
(-9646 in your data structure,-9646,1300), so it appears that ti is
the function that is going wrong, and in unpredictable ways - for
example doing this again gives a different error on a different
number.
The ti function looks incredibly simple, and I can only guess that
there is some memory corruption going on? Something like that, as the
program does different things on different executions.
Unfortunately I couldn't find any repeatable specific test case, but
the above check function is pretty much guaranteed to get it wrong
every time. This is using May 2006.
Thanks
Neil
On 9/22/06, Hugs
#37: Equal Ints are not equal ----------------------+----------------------------------------------------- Reporter: guest | Owner: nobody Type: defect | Status: new Priority: blocker | Milestone: Component: hugs | Version: current Resolution: | Keywords: ----------------------+----------------------------------------------------- Comment (by guest):
{{{ Hi,
a few days ago I wrote a program which handles my own integers. Then I check my own implementation against the normal Int implementation in Hugs. Thereby I wonder about inequality of Ints which are definitely equal.
This is a test program I wrote after I recognized that hugs has probably a bug in the inequality check of Ints.
data MInt = Zero | Succ MInt | Pred MInt deriving Show
tn :: Int -> MInt tn x | x<0 = Pred (tn (x+1)) tn 0 = Zero tn (n+1) = Succ (tn n)
ti :: MInt -> Int ti Zero = 0 ti (Succ x) = 1+(ti x) ti (Pred x) = (ti x) -1
testi :: (MInt -> MInt -> MInt) -> (Int -> Int -> Int) -> Int -> Int -> Bool testi f g x y = (ti (f (tn x) (tn y))) /= (g x y)
myMul x y = tn ((ti x) * (ti y))
test = [(x,y,ti (myMul (tn x) (tn y)),x * y)| x<-[-100..100], y<-([-100..(-1)]++[1..100]),(testi myMul (*) x y) ]
the list "test" should be empty in any case but unfortunately hugs returns this (copy of the console):
__ __ __ __ ____ ___ _________________________________________ || || || || || || ||__ Hugs 98: Based on the Haskell 98 standard ||___|| ||__|| ||__|| __|| Copyright (c) 1994-2003 ||---|| ___|| World Wide Web: http://haskell.org/hugs || || Report bugs to: hugs-bugs@haskell.org || || Version: November 2003 _________________________________________
Haskell 98 mode: Restart with command line option -98 to enable extensions
Type :? for help Main> test [(-100,80,-8000,-8000),(-100,99,-9900,-9900), (-99,58,-5742,-5742),(-99,83,-8217,-8217), (-98,71,-6958,-6958),(-98,86,-8428,-8428), (-97,46,-4462,-4462),(-97,87,-8439,-8439), (-96,76,-7296,-7296),(-95,88,-8360,-8360),(-94,73,-6862,-6862) ERROR - Control stack overflow Main>
I have checked this program with Hugs on different plattforms: Gentoo Linux x86 64-Bit and Suse Linux 32-Bit. On both plattforms the same bug occurs with different Ints for each run (i.e. the list contains different tuples). I have also checked my program with other Haskell implementations and there it returns an empty list as expected.
Best regards, Stephan Swiderski (swiderski@informatik.rwth-aachen.de)
P.S.: We have tested Sep2006 with similar results: __ __ __ __ ____ ___ _________________________________________ || || || || || || ||__ Hugs 98: Based on the Haskell 98 standard ||___|| ||__|| ||__|| __|| Copyright (c) 1994-2005 ||---|| ___|| World Wide Web: http://haskell.org/hugs || || Bugs: http://hackage.haskell.org/trac/hugs || || Version: September 2006 _________________________________________
Haskell 98 mode: Restart with command line option -98 to enable extensions
Type :? for help Main> :e Main> test [(-100,90,-9000,-9000),(-98,40,-3920,-3920),(-97,63,-6111,-6111),(-97,69,-6693,- 6693) ERROR - Control stack overflow
}}}
-- Ticket URL: http://hackage.haskell.org/trac/hugs/ticket/37 Hugs http://www.haskell.org/hugs/ Hugs 98, an interpreter for Haskell _______________________________________________ Hugs-Bugs mailing list Hugs-Bugs@haskell.org http://www.haskell.org/mailman/listinfo/hugs-bugs