My 50 solution for grid computer should fail a more rigorous testsuite.

A case where two lines are the same will produce the wrong result, but my program still passes.

e.g. (say for a 3x3)

1 1 1
1 1 1
1 1 1

my program would return 9, because of the way I'm identifying each line of input.

Hi Jasper. This is not like the good old Perl golf days with a human referee and golfers pointing out bugs in the test program, which the referee then remedies, thus disqualifying invalid entries. Once a codegolf game gets started, it seems there is no going back; it's all automatic and there is no way to fix a poor test program. Grid golf has a poor test program in that it just generates twenty random grids (see the Forum for this game for other examples of faulty entries that "passed"). For an example of a much better test program, see the Roman to Decimal golf, which has a fixed number of nasty test cases in addition to some random ones -- though a more rigorous one would test all 3999 Roman numerals, doing that that would break the four second time limit. I've formed this conclusion because I noticed in http://codegolf.com/boards/conversation/view/98 that carldr (this site's owner) answered a question with "Yep, once passed, always passed! Milk holes in the tests as much as you want. " Personally, I dislike this aspect of codegolf.

ergh, that's pretty unpleasant, in that repeatedly submitting the same solution could get it approved at some point.

Oh well.