Ruleset for tests

[bormilan]

I'm working with Elvis on a lot of different repos, and I realized that there are some rules that are most of the time need to be set to ignored in the test files group.
For example, the dont_repeat_yourself. I think it's not a bad practice to write tests that violate DRY because it's better to keep the tests as readable, understandable, and simple as possible.

My idea is to create a ruleset specifically designed for test files.

Also, if we make something like this, we should consider making default rule configs different for tests than the code files. (like set max_function_length higher)

[elbrujohalcon]

Yes!!! I would definitely use it, too!!

[paulo-ferraz-oliveira]

I think categories are about grouping rules into their intrinsic meaning, but if you wanna have different defaults we have to find a way to differentiate them when "inside tests" or "outside tests". I think a pre-defined ruleset with some values for testing (defaults) might make sense, but I haven't thought it through.

Yeah, ignore, whatever is the case, should always be applied last. So what you refer to makes sense (strict - refactoring).

My initial goal with categories, though, will probably just be around classification so you can choose how to group results, not classification on how to run rules (a good idea, but I hadn't thought so far ahead, yet).

[elbrujohalcon]

I would start very very simple: Just a ruleset for tests, with the same rules as erl_files except for dont_repeat_yourself. That would be my first PR, for sure.

Then we might consider allowing to set specific defaults for each rule on each ruleset. I think this will not be trivial mostly because it has to be easy to understand for both regular users (which will just pick a ruleset without much consideration of the implications) and ruleset designers (i.e., advanced users that define their own rulesets).

[bormilan]

I like it, I made the first issue: #435

[paulo-ferraz-oliveira]

Another alternative would be to allow us to define custom rulesets (if that's not already possible) like erl_files - {rule, dont_repeat_yourself} which I guess would be easier to maintain (a generic mechanism) than yet another ruleset. Just food for thought, though...

[elbrujohalcon]

Let's follow this conversation in #436 ⏩

[paulo-ferraz-oliveira]

Closing in favour of #436.