Linter: The non_constant_identifier_names lint should ignore a leading underscore

[devoncarew]

The non_constant_identifier_names should likely ignore a leading underscore. For methods names like void _checkAssignment_recursion(), I see:

info: Name non-constant identifiers using lowerCamelCase. (non_constant_identifier_names at [nnbd_migration] lib/src/edge_builder.dart:3313)

Also, the code for the lint is non_constant_identifier_names, which seems like it describes the items it applies to, instead of what it does (like 'prefer_lower_camel_case').

[bwilkerson]

I agree that a leading underscore should be ignored.

Side note on naming: Yes, it's unfortunate that the lint names don't all conform to a standard naming scheme. If it weren't for the pain it would cause users I'd suggest renaming them all more consistently, but I can't convince myself that it would be a net positive from a user perspective. But we realized awhile back that we'd made a mistake by using "prefer", "avoid" and other terms from the style guide because the style guide has, in a few cases, changed since the lints were originally named and now the names are inconsistent.

[pq]

The issue w/ lint naming inconsistency bothers me daily. I really wish we'd been more principled and consistent but as Brian points out, our ideas have evolved. Thinking ahead, it'd be cool to do a renaming at some point but so far the pain hasn't seemed worth it. (Maybe we can revisit if bulk fixing takes off?) (FWIW there's some related discussion in #1179.)

As for this lint misfiring, my sense is that it should be further expanded to allow for underscores in function declarations too (and am somewhat baffled that it hasn't already...) Thoughts?

[devoncarew]

If there are a few lint names that really stand out, we could rename them to get a better sense for the cost to the ecosystem. We'd likely want to do something like have two names for the lint - the new and the deprecated old - show a deprecation messages, and possibly, have a quick fix to convert from the old to the new.

[bwilkerson]

That sounds like a reasonable path forward. Perhaps Phil and I can carve out a few hours to review the lint names at some point to see what the magnitude of the changes might be.

[pq]

SGTM!

[pq]

Taking a closer look, the issue here is not the leading underscores, it's what follows. Specifically _recursion(). To fix it, consider renaming to: _checkAssignmentRecursion().

[pq]

Closing as working as intended.