in reply to Working around a 5.6.0 bug
This craving people have for perfect test results (and sometimes more absurdly, 100% coverage) is excessive. Tests are great for many reasons, but they are forever A TOOL.
Do you expect your code to fail on 5.6.0? If not, you shouldn't have this test. Not "shouldn't have this test fail", but not have this test, period, because it isn't indicative or a real problem. If your code can fail on 5.6.0, either make 5.6.1 a prerequisite, or decide to fix the problem -- in the code, first -- knowingly bloating your code but happy in the knowledge that you know why you did it (and that tests helped you make an informed decision).
|
|---|
| Replies are listed 'Best First'. | |
|---|---|
|
Re^2: Working around a 5.6.0 bug
by Anonymous Monk on Dec 09, 2004 at 21:44 UTC | |
by gaal (Parson) on Dec 10, 2004 at 06:13 UTC |