Unused instance and local variable accesses

[nobodywasishere]

I've been developing new rules for ameba, and as part of that have been testing them on this repository to verify their functionality. I've found a few issues that could be fixed.

src/compiler/crystal/tools/formatter.cr:

src/compiler/crystal/syntax/parser.cr:

src/compiler/crystal/semantic/filters.cr:

There are several others, but they're mostly having the var on the last line of a method that has a Nil return.

[straight-shoota]

Happy to accept a patch.

I'm not sure it makes much sense to track such individual linter issues in a ticket.
As long as the code works, it's fine. Of course we'd want to improve the quality.
But IMO it's not important enough to keep a bookmark for every detail. Especially considering that they should be relatively straightforward to just fix.
We could have a meta issue to keep track of linter problems, though.

[nobodywasishere]

I agree, I'll start working on an initial PR to start fixing these and other issues. The check_end one concerns me though, as it may introduce a breaking change by fixing it. I'm fine turning this into a ticket to keep track of linter issues / discuss them

[straight-shoota]

Related:

• Enable ameba in this repo #14608

[straight-shoota]

Of course, if there's a problematic detail to discuss on a specific linter issue, such a discussion is much appreciated.

It wasn't clear to me that the check_end variable is shadowing a method. So yeah we ought to be careful about that. Apparently the formatter specs are not dense enough to catch that these check_end are not actually checking the end.