unnecessary_statements and no_self_assignments discussion

[FMorschel]

If you have a code like the following we should be receiving unnecessary_statements triggers:

void f(int a) {
  a = a;  // <-- No lint
}

The background for this is that while working on #57090 I ended with a code similar to:

int i;
if (other case int? i?) {  // It was an actual case but this is not the point
  i = i;   // shadowed i self-assigning 
} else {
  return;
}
i; // <-- The non-nullable local variable 'i' must be assigned before it can be used.

In my case the variable in question was even the same name as a propriety for the class case I was testing. So by doing like MyClass(: var i) I completely shadowed my variable and was left wondering why I got that error bellow.

[dart-github-bot]

Summary: Self-assignment (a = a) doesn't trigger unnecessary_statements, but should. This impacts shadowed variables, causing confusion with null safety.

[bwilkerson]

That's valid for local variables (including parameters), but isn't safe to conclude for getters and setters, both of which can have side-effects. They shouldn't have side-effects, but they can, so we can't assume that they won't.

[parlough]

Note that there is a separate no_self_assignments rule. Is that sufficient?

[FMorschel]

Yes, I think it would be. Then two questions:

• What about the concerns that @bwilkerson raised? The examples on the site do suggest it would warn for getters-setters.
• Why aren't these on the recommended set? It feels like they are simple enough that they should.

[bwilkerson]

Why aren't these on the recommended set?

I don't know, but my guess is that it's probably because the lint flags the use of getters and setters, and as a result there was probably a feeling that there's too much potential for false positives.