On Mon, Dec 14, 2009 at 12:43:16PM -0800, Robert Miller wrote: > > This is an interesting example -- it turns out *not* to be a bug. The > matrix printed takes about 725 seconds, but when you apply the > permutation printed, you get an isomorphic matrix which takes 5000 > seconds. I suppose this is further support for removing these tests, > since it is just too hard to predict the runtime. >
Since I know nothing about the algorithm, I couldn't tell whether the large differences in runtime between different runs was normal or a sign of trouble. The question that I was asking myself while waiting for the tests to end was this: how do I give you a good bug report? If the computation ends and the checks don't work out, then the testing code outputs the matrix and permutation, so I just have to send you that. But if I am suspecting that the computation is stuck, or maybe it's just unnecessarily long, what can I do to find out what input it was working on so I can pass it along? Maybe trapping Ctrl-C and having the testing code say something like: Computation interrupted while working on matrix = permutation = Then I could send that to you to check. (Or if I got myself better informed, check it myself.) Also, when we'll have a battery of randomized tests running 24/7 on some machine hunting for bugs, putting in a timeout for these tests and reporting the input when this timeout is exceeded might help. (Looking on the bright side, your code has been tested quite extensively so far by everybody running Sage's doctests, and we seem to not have found any bugs yet, so that's quite something!) Best, Alex -- Alex Ghitza -- Lecturer in Mathematics -- The University of Melbourne -- Australia -- http://www.ms.unimelb.edu.au/~aghitza/ -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
