Level 2: If there is a bug, it will result in a runtime error describing the problem. This case is worse than the previous one in a couple of ways.
- First, it takes more steps to discover a bug. The cycle of edit-compile must be lengthened to edit-compile-run-test. This will slow down your debugging quite a bit.
- Second, a bug may be in a piece of code that does not get executed every time the program is run, or it may only occur under special circumstances. A lot of testing could be required before it is discovered, and even then, it is hard to be confident that all bugs have been found.
This is the reason that languages with static typechecking, such as Java, are generally preferred when writing large applications. In a dynamically typed language such as Python or Ruby, you can invoke any method on any object. The compiler does not know what type of object is stored in any variable, and just trusts that it has a method with the specified name that takes the right set of arguments. If this assumption is wrong, the result is a runtime error.
To be fair, this can be a very useful tool in certain cases. It also allows your source code to be slightly more concise, which is convenient when writing simple scripts. But for large applications, those benefits are far outweighed by the lack of compile-time error checking. This is one of the reasons that so many more applications are written in Java than in Python.
A level 2 bug opportunity is definitely worse than a level 1, but it is still not that bad. If all bugs were of this sort, debugging would be a relatively simple and painless process. Now let's consider a case that really is bad.