#8695: Arithmetic overflow from (minBound :: Int) `quot` (-1) ------------------------------------------------+-------------------------- Reporter: rleslie | Owner: Type: bug | Status: patch Priority: normal | Milestone: 7.8.1 Component: libraries/haskell2010 | Version: 7.6.3 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Incorrect result at runtime | Unknown/Multiple Test Case: | Difficulty: Blocking: | Unknown | Blocked By: | Related Tickets: #1042 ------------------------------------------------+-------------------------- Comment (by rwbarton): Choosing the behavior to match the existing documentation is the wrong way around, in my opinion. If the current behavior is considered desirable, then there is a documentation bug, not a bug in the definition of division. Unfortunately, there is no choice of behavior that is best in all contexts. For the kinds of programs I most often write, which are computations I run on my own computer, raising an exception is preferable ("safer"). I'm likely to not consider the possibility of dividing `MIN_INT` by `-1`, and the exception protects me: If my program runs to completion, then I know the answer is correct (and that such a division never occurred). In the unlikely event that I do get an exception, then I can consider what to do in the event of such a division and ensure that it is treated properly. If the division silently produces `MIN_INT` then there's some chance that this was the correct thing to do in my context, and some chance that I silently got a nonsense answer, which to me is the worst possible outcome. On the other hand, for someone who is writing a server to, say, evaluate arithmetic expressions, it is a pain if a malicious user can crash the server by asking it to evaluate `MIN_INT / (-1)`, and there's not really any sense in which answers are "right" or "wrong" anyways. For a program that does division and then does something with the result (like indexing into an array) I tend to worry about the security implications of working with a result that may not be expected by the developer (though something has likely gone wrong already, in the specific event that we are dividing `MIN_INT` by `-1`). Personally I am against this change because it would make ghc less useful for me. Moreover, I tend to feel that servers are complicated bits of software that are prone to crashing for many other reasons already, and it is prudent to ensure that an exception raised while handling one request cannot bring down the whole server anyways. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/8695#comment:14 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler