Our rubocop config hides all current violations. It allows us to have a
passing rubocop run on the current code and improve it gradually. It
detects new violations, but doesn't annoy us with all the existing ones.
Code Climate has its own way of remembering all current violations which
is more sophisticated than ours. The new config for Code Climate doesn't
hide any violations so that Code Climate can give us a realistic score
of code quality and warn us about every new violation.
Splitting the configuration into the main three parts gives a quicker
overview and makes maintaining the parts easier.
One of the biggest pros of linters like Rubocop is to get valuable feedback to
help write better code. The way we have Rubocop configured now we don't prevent
new code from adhere improved code quality and this is specially important when
touching code that already suffers from complexity.
Without all Rubocop's Metrics cops enabled there's no way to get this insights
and write better code. This enables them while regenerating the
`.rubocop_todo.yml` to hide the current violations.
So, next time we touch existing code that we think could be simpler, we should
go to `.rubocop_todo.yml` and remove any occurrences of the file in question.
This way we could Rubocop's feedback right in the editor. This is tremendously
helpful when refactoring. It shows you where to start.
There's no point of using Code Climate if most of its checks are
disabled. Now it scores our codebase with an A and reports just 2 code
smells, 64 duplications and 14 other issues. That couldn't farther from
reality. If we care about the architecture and maintainability of our
app, this is a must.
To transition towards enabling all default checks, I started with
scss-lint, duplication and method-complexity. This doesn't mean will
have to fix them all straight away as we'll add the current violations
to the `.rubocop_todo.yml` but it'll point us the parts where we should
focus our efforts.
