Barry Hart

PR #3697 avoids the looping behavior. Lint issues are still flagged, but the rule does not attempt to fix it _if_ it would cause a loop. We should still try...

The plan sounds good overall. I had to skim it just now, but one thought I've had about this is that the depth of a segment in the parse tree...

The plan sounds good overall. I had to skim it just now, but one thought I've had about this is that the depth of a segment in the parse tree...

Can you provide a full example query? Also, what dialect are you using above?

It sounds good to me. I could see this as a new rule out a new, optional behavior for L003. (Either seems okay from a user perspective. I'm not sure...

>I think a new rule would be simplest That's my guess, too. If necessary, you can copy or otherwise reuse some code from L003. We have a few other rules...

We'll want some tests for that scenario. Idea: Copy several of the existing L003 tests and make the existing query a subquery.

Good luck! I've had some struggles with understanding L003 as well -- it's probably the most complex of the linting rules (at least that I've touched). As you're looking at...

@alanmcruickshank: I'm okay, thanks! This is a slow-moving, slightly experimental PR anyway.

@yajo: This is ready for review. Most of the code is copied from #627; I mainly organized things a bit differently to support both old and new behavior. As a...