Re: [html-tests] Tests for navigating window.opener across origins (#2356)

Tests look fine to me: the first two pass in Chrome, the last does not because of the logic in https://cs.chromium.org/chromium/src/third_party/WebKit/Source/core/frame/Frame.cpp?rcl=56b1ac1be5b34735541c864d6de5b61e3091258d&l=347 which enforces a direct relationship between the window initiating the navigation and the window being navigated (same-origin, `parent`, or `opener`).

Given the reference to @abarth, et al's 2008 paper in that part of the code, I assume it simply predates the current definition in HTML. As you note, the frame can simply navigate its parent to a page that can navigate `opener`, so we should probably change Blink to align with Gecko.

View on GitHub: https://github.com/w3c/web-platform-tests/pull/2356#issuecomment-287302014

Received on Friday, 17 March 2017 09:04:34 UTC