fixes #11474

[Araq]

No description provided.

[timotheecour]

=> #11773

[krux02]

I have to support the criticism of @zah here.

• no documentation
• no specification
• no tests for error messages
• no review process (did we ever even consider if using the [] operator here could introduce ambiguities in the Nim grammar?)

Introducing new features into Nim like this can really Hurt the language. Please revert this commit and make a proper RFC for it with a specification of the behavior. And then the PR must contain tests for that ensure the misuses of unref will trigger proper error messages, instead of doing undefined behavior.

[dom96]

No need to revert it, but discussion should be had, I agree with @krux02 and @zah on that.

[Araq]

did we ever even consider if using the [] operator here could introduce ambiguities in the Nim grammar?

How could it? The parser was left untouched. But I completely agree with all your other points.

[krux02]

@Araq since this PR got removed again, I think it is not that relevant anymore but

How could it? The parser was left untouched. But I completely agree with all your other points.

An nkBracketExpr on a type is usually a generic invocation. That is the ambiguity I am talking about. And my point is not if it is ambiguous, but that we should at least have a review process to see if there could be an ambiguity.