Supply status for missing exit codes

[CyberiaResurrection]

Following on from the work in "Handle -9 PyTest return code", discussion with @Otto-AA and @boxed revealed it would be desirable to handle missing return codes in full generality.

The simplest way I could think of to do that, trying to eliminate side effects, is to convert status_by_exit_code to a defaultdict, with a default value of "suspicious" - one already defined, and handled, elsewhere in the codebase.

[Otto-AA]

Simple, but should work fine. You could test it by removing some other exit codes from the dict (eg the codes for killed) and then checking if it outputs suspicious instead of killed with the mutmut browse command.

[Otto-AA]

You likely also have to make exit_code_to_emoji a defaultdict too

[Otto-AA]

I think this line should use a direct dict lookup instead of the .get method (or simply exit_code_to_emoji):

mutmut/mutmut/__main__.py

Line 1127 in 229db0f

print(emoji_by_status.get(status_by_exit_code.get(exit_code), '?'), mutant_name)

I think with the .get() it won't use the defaultdict

[CyberiaResurrection]

You likely also have to make exit_code_to_emoji a defaultdict too

If the defaultdict value for status_by_exit_code wasn't already in exit_code_to_emoji, yes, definitely. As I've gone with "suspicious", which is already in exit_code_to_emoji , I'm not (at a surface level) seeing why it should be a defaultdict.

[CyberiaResurrection]

Simple, but should work fine. You could test it by removing some other exit codes from the dict (eg the codes for killed) and then checking if it outputs suspicious instead of killed with the mutmut browse command.

Yes, it does - removing the exit code status for -9 results in that particular mutant being logged as "suspicious" instead of "segfault".

[CyberiaResurrection]

I think this line should use a direct dict lookup instead of the .get method (or simply exit_code_to_emoji):

mutmut/mutmut/__main__.py

Line 1127 in 229db0f

print(emoji_by_status.get(status_by_exit_code.get(exit_code), '?'), mutant_name)

I think with the .get() it won't use the defaultdict

According to https://docs.python.org/3/library/collections.html#collections.defaultdict , your reading of get() behaviour is correct.
On other hand, using the defaultdict lookup is a behaviour change above and beyond what I set out to do in this PR.
On third hand, the ? method (not checked - ie, no tests run) when receiving the result of a test run with a hitherto-unseen exit code could come across as misleading, as at least one test has actually run for the mutant in question.

Sod it, after writing that, I'll make the change you suggested.

[Otto-AA]

We have following definition:

exit_code_to_emoji = {
    exit_code: emoji_by_status[status]
    for exit_code, status in status_by_exit_code.items()
}

This uses all items in status_by_exit_code to define exit_code_to_emoji. However, this does not copy the defaultdict behaviour. exit_code_to_emoji is still a normal dictionary and not a defaultdict, so it will fail when we access a non existing key.

Thus, I would suggest to make it a defaultdict. Though it's interesting that your testing succeeds even though this is not currently implemented yet.

[boxed]

Using .get is an alternative to a defaultdict. It won't raise and can be supplied a default value.

The advantages of a normal dict are that it's nicer to write dict literals, and that "new" keys don't take up space as they do if you use a defaultdict. In this case that doesn't matter much.

[CyberiaResurrection]

@boxed , @Otto-AA , any objections to me merging this one?

[boxed]

None. Go ahead.