Open szmate1618 opened 2 years ago
Ok, I did not expect those test failures. I did run the tests on my machine, at least on Windows, and they passed (except for the ones that are skipped on that platform). edit: I accidentally ran the tests on an earlier version, without my changes.
I will look into this tomorrow.
Fixes #162.
This pull request currently contains 2 changes:
croak
on empty (or undefined) strings in_search_path
croak
on references to undefined arrays inharness
Any of the two would be sufficient on its own to solve this specific problem, but I suggest to go with both.
I think the first change is important because regardless of this issue, it is semantically incorrect to search for empty file names on a path. And the second change is important because if we know we are going to fail, it's better to fail fast. Of course, failing really fast would mean the very first line of
run
, but after studying the code a little, it felt like the correct place to handle such special cases is inharness
.Instead of failing silently I'm using
croak
in both cases, but I'm unsure if this is the correct decision, and would like to hear your input. I'm unsure, because this library has many users, and it is quite possible that someone has accidentally (or not) written code that depends onrun
not failing on undefined values. This fix would break such code. On the other hand, if your code accidentally depends on a bug, maybe you want it to break, so you notice the problem.I would add some tests, but to be perfectly honest, I have no idea how the test framework works.