Intent to implement and ship: disown opener in a targeted navigation when noopener is used

88 views
Skip to first unread message

Lucas Gadani

unread,
Nov 3, 2017, 11:15:28 AM11/3/17
to blink-dev

l...@chromium.org https://github.com/whatwg/html/issues/1826 When targeting links with rel=noopener the opener should always be disowned. That means this: <a href='' target='foo'>Click me first</a> <a href='' target='foo' rel='noopener'>Click me second</a> will open two different windows. The same behavior is expected when using window.open('', 'foo', 'noopener'). Disowning the opener allows the browser to make better decisions about whether to launch a new renderer process for the new window.


We also want to achieve interoperability with other browsers when it comes to the 'noopener' behavior.


Firefox: Shipped Edge: No public signals Safari: No public signals Web developers: No signals

Firefox initially shipped an implementation that disowned both 'noopener' and 'noreferrer' but they found that disowning 'noreferrer' was not web-compatible (https://bugzilla.mozilla.org/show_bug.cgi?id=1358469).


The intent here is to implement this only for 'noopener'. Firefox ships this since 52.


Yes.

https://crbug.com/651661 https://www.chromestatus.com/features/5641309343711232 Yes.


Rick Byers

unread,
Nov 3, 2017, 1:02:14 PM11/3/17
to Lucas Gadani, blink-dev, Domenic Denicola, Mike West
Sounds reasonable to me, just a couple questions:

Do we have any stats on how often "noopener" is used, to form an upper bound on the potential for breakage?  Mike, what's your judgement on the compat risk?

What's the plan for getting the spec and web-platform-tests updated for this?

--
You received this message because you are subscribed to the Google Groups "blink-dev" group.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CALQVofrmUM3Nq8S2%2BjXdUn3_BoYR10D2HhTUKdxBeURToo0K3Q%40mail.gmail.com.

Boris Zbarsky

unread,
Nov 3, 2017, 1:50:03 PM11/3/17
to Rick Byers, blink-dev, Mike West, Domenic Denicola
On 11/3/17 1:01 PM, Rick Byers wrote:
> What's the plan for getting the spec and web-platform-tests updated for
> this?

There are some tests for this already. See
http://w3c-test.org/html/browsers/the-window-object/window-open-noopener.html
and
http://w3c-test.org/html/semantics/links/links-created-by-a-and-area-elements/htmlanchorelement_noopener.html

As for the spec getting updated... it's somewhat waiting on you guys
saying whether you're actually on board with the specific proposals
involved and explaining what it is, exactly, you're planning to
implement. See https://github.com/whatwg/html/issues/1826 where I've
been waiting for a response for the last 6 months or so.

-Boris

Philip Jägenstedt

unread,
Nov 15, 2017, 8:08:18 AM11/15/17
to Boris Zbarsky, Rick Byers, blink-dev, Mike West, Domenic Denicola
Lucas, can you work on getting https://github.com/whatwg/html/issues/1826 resolved, with tests, matching what it is you'd like to implement and ship here? If there's some part of it that is in doubt and needs to be proven web compatible before being changed in the spec, tentative tests may be of use.

--
You received this message because you are subscribed to the Google Groups "blink-dev" group.
Reply all
Reply to author
Forward
0 new messages