proposal: `null_aware_followed_by_null_assert`

[srawlins]

null_aware_followed_by_null_assert

Description

This one is a little odd, but I've seen a few cases. In code like this:

    foobarDateRange = (foobarDateRange
          .clampSomething(Something.VALUE)!
          .applySomething()
          ?.clampToSomething(Something.OTHER_VALUE)!)!;

It's a bit hard to follow the parens and bangs, but that last line essentially has ?.foo(bar)! in it. I don't think there is ever any reason to have a null-aware method call or null aware property access (or chain of these) followed by a null-assert. It fails in the same way as !.foo(bar), with a more-or-equally informative stack trace, and is more readable.

Kind

Readability

Bad Examples

a.b?.c!;
a.b?.c()!;
a.b?.c.d().e!;

Not sure if this needs to be careful around cascades...

Good Examples

Add a few examples that demonstrate a “good” adoption of this lint’s principle.

[bwilkerson]

What about cases where the chain after the ? can also introduce a null? For example, if foo can return null, is ?.foo(bar)! better than !.foo(bar)!?

[srawlins]

Right. It is not. A null-assert that throws an exception in ?.foo(bar)! is either thrown because foo() returned null or because null-shortening from the ?. got us to the null-assert.

If the second null-assert throws an exception in !.foo(bar)!, it is because foo() returned null. If the first throws an exception, it is because the value to the left of the code in question returned null.

So the latter carries more information.

[dart-github-bot]

Summary: User proposes a lint to flag null-aware operations followed by null assertions. This pattern is considered redundant and less readable than a direct null assertion.