- From: bifurcation <notifications@github.com>
- Date: Thu, 12 Nov 2015 13:49:42 -0800
- To: heycam/webidl <webidl@noreply.github.com>
- Message-ID: <heycam/webidl/pull/65/c156246766@github.com>
Assuming we're going with the "throw/reject" approach and not the "exposure" approach, this patch looks sensible to me. In an ideal world, I would really prefer we control this through exposure rather than errors, for reasons I've stated before. Is it possible to have use exposure as a general approach and still be consistent with w3c/webappsec-secure-contexts#10? The case you have to deal with is: 1. `if (navigator.secureOnlyFeature) {...}` 2. `document.domain = document.domain` 3. `navigator.secureOnlyFeature.doSomething()` What if we were to break the symmetry of w3c/webappsec-secure-contexts#10, in the following way: If you touch something secure first, then `document.domain` throws. If you touch `document.domain` first, then you can't see anything that's `[SecureOnly]`. That's a little bit magical, in that the `[SecureOnly]` stuff gets yanked from the DOM when you touch `document.domain`. But it's not clear to me that hiding all of those things is any more work than telling them all to throw. --- Reply to this email directly or view it on GitHub: https://github.com/heycam/webidl/pull/65#issuecomment-156246766
Received on Thursday, 12 November 2015 21:50:12 UTC