Concretely, the problem was that the variable def-use lint treats there exists [...] _X_ as declaring _X_, which is true for uses like there exists an integer _X_ such that or there exists an element _A_ of _B_, but not for there exists a value in _X_, which is the form in used in the relevant line in the PR.
Specifically, the line in 262 is If there exists a CharSetElement in _A_ containing exactly one character _a_ such that [...], which should use_A_ but declare_a_; prior to this PR it was treating both as declarations.
Should unblock https://github.com/tc39/ecma262/pull/3102.
Concretely, the problem was that the variable def-use lint treats
there exists [...] _X_as declaring_X_, which is true for uses likethere exists an integer _X_ such thatorthere exists an element _A_ of _B_, but not forthere exists a value in _X_, which is the form in used in the relevant line in the PR.Specifically, the line in 262 is
If there exists a CharSetElement in _A_ containing exactly one character _a_ such that [...], which should use_A_but declare_a_; prior to this PR it was treating both as declarations.The linter wasn't tripping earlier because a separate issue in ecma262 (introduced in https://github.com/tc39/ecma262/pull/2418 and fixed in https://github.com/tc39/ecma262/pull/3102) caused
_A_not to be captured, which basically canceled out the bug here.