On Thu, Apr 21, 2016 at 10:50:14PM +0200, Joachim Breitner wrote:
I don’t think it is entirely unrelated, but there is a separate ticket about this: https://ghc.haskell.org/trac/ghc/ticket/10246 It even proposes a fix.
But if you want to be unblocked in implementing #11297, simply comment out the assertion in Literal.hs (just for you), or compile without -DDEBUG.
Without asserts it's reproducing. Thanks a lot!
Is there anything I can do with this? Is it because that test has 64-bit unboxed integers and I'm running it on 32-bit VM?
Right, but that is the point of #11297: The test case is tailored to 64 bit, and the task is to create a second test case with 32 bit numbers.
Yes, I saw that it's tricking integer overflow but only for 64 bit. I will come back later with the new test. Thanks, Alex