[`unnecessary_parenthesis`] false negative in conditional operator

[FMorschel]

When doing this:

final email = (user.email != null) ? (user.email!) : '';

This is a dumb example, but my point is that in other ternary expressions that I've used, I saw this.

I think that the (user.email!) expression should trigger this as well. I'm not sure because this might be intentional to leave ternary expressions with possible parentheses around all possible results and the test, but it seems strange to me that it doesn't warn me when that is not strictly necessary.

This is a false negative that occurs in any of the three places of a ternary operator.

_Originally posted by @FMorschel in #57137

Edit: here is the CL: https://dart-review.googlesource.com/c/sdk/+/390161

Edit 2: Linking the related issue: #59069

[lrhn]

It does appear there is an exception for the brances of conditional expressions.
Maybe that exception doesn't need to apply to all expressions. A primary + selectorChain may not need parentheses. Maybe suffix/prefix increment operators can be included too, so c ? foo.bar.baz[42].qux++ : x++ won't allow parentheses.

I usually only want parentheses when the syntax contains whitespace, so something with binary operators or prefix operators that are words, fx await. Prefix/suffix operators are flush against the expression, and selector chains only contain spaces inside argument lists, type argument lists and string literals, which are already delimited. Could consider not exempting those from the unnecessary parentheses lint.

(Pet-quixotic-peeve: The ?/: is the conditional operator/expression. It's sometimes called "the ternary conditional operator" or "ternary if", but it's imprecise to just say "the ternary operator". It's a ternary operator, but []= is also a ternary operator in Dart - it takes three operands. This is likely a lost cause, but by golly, I'm going to keep trying!)

[srawlins]

Similar to #57137, this is a case where the rule does not report the parentheses, because they can improve readability. See the test case, test_notEqualInside_returnExpression (return (1 != 1);), which has parens that are not reported.

[FMorschel]

@srawlins, yes, I do agree that they can make readability better. I just disagree that the following has no warnings:

(list.isEmpty) ? (false) : (false);

Neither of the three has spaces or would be more readable because of the parentheses.

[srawlins]

Ah I missed the user.email! case.

I just tested at head, and one lint diagnostic is reported for:

class U {
  String? email = '';
}
var u = U();
var x = 1 == 2 ? (u.email!) : false;

but none are reported for:

var list = [];
var x = (list.isEmpty) ? true : false;

I agree something should be reported for the second case.

[FMorschel]

I've downloaded the latest sdk minutes ago (from here) and I got no lints in:

class U {
  String? email = '';
}
var u = U();
var list = [];
var x = 1 == 2 ? (u.email!) : (list.isEmpty);

I'm not sure what you got. I do have unnecessary_parenthesis set to true.

---

Edit

I tested with:

(list.isEmpty) ? (false) : (true) ? (0) : (1);

And all (false), (true), (0) and (1) triggered.

[FMorschel]

I've opened https://dart-review.googlesource.com/c/sdk/+/390161