Figure out best way to work with the AST in linter rules about regal HOT 2 CLOSED

I too think that 3 is appealing. I have done something like 2 in the past to reduce the size of the input needed and it worked ok. I think that the main benefit of 3 is that the rule could make use of what it needs, rather than needing to declare what it needs upfront.

from regal.

Closing this as done, as we've largely settled on option #3 here. There are some limitations to the approach where we'll still need to perform multiple walks to satisfy some rule, but it's probably the best we can do for now.

If OPA ever gets support for dynamically invoked functions, we could investigate the cost of doing a single walk across the AST, and have each rule be a function that got invoked only when it's conditions were met, like "check prefer_snake_case whenever a variable declaration is encountered"... but it'd require that function invocations are cheap enough, and I'm not sure that would be the case. Doing this today with rules and with makes linting take minutes rather than seconds.

from regal.