On Wed, Feb 9, 2022 at 10:50 AM Yoav Weiss <yoavwe...@chromium.org> wrote:
> > > On Wednesday, February 9, 2022 at 2:49:55 AM UTC+1 Mike Taylor wrote: > >> Hey Xiaocheng, >> >> Thanks for working on improving interop! A few thoughts and questions >> below. >> >> On 2/8/22 7:25 PM, Xiaocheng Hu wrote: >> >> Contact emails xiaoche...@chromium.org >> >> Explainer None >> >> Specification None. Not a standard feature. >> >> Summary >> >> Event.path is a non-standard API that returns the event's path, which is >> an array of the objects on which listeners will be invoked. It is supported >> by Blink only, causing web compatibility issues. Web developers should >> switch to the equivalent standard API Event.composedPath(), which returns >> the same result. >> >> >> Blink component Blink>DOM >> <https://bugs.chromium.org/p/chromium/issues/list?q=component:Blink%3EDOM> >> >> TAG review >> >> TAG review status Not applicable >> >> Risks >> >> >> Interoperability and Compatibility >> >> The removal of this API should improve interoperability, as it's >> supported by Blink only. It still has 18% usage as of Feb 2022 ( >> https://chromestatus.com/metrics/feature/timeline/popularity/345), so we >> will only deprecate it for now, and will not remove it before the usage >> drops low enough. We expect low compatibility risks, since there is an >> equivalent standard API (Event.composedPath()) by all browsers, and the >> following polyfill should also keep existing sites functioning with minimum >> changes: >> >> 18% is a _lot_ of usage. So much that I'm surprised there aren't dozens >> of compat bugs reported against Firefox. In >> https://bugzilla.mozilla.org/show_bug.cgi?id=1697590#c0 there are only 2 >> linked site bugs. And there's only 3 in >> https://github.com/webcompat/web-bugs/issues?q=is%3Aissue+composedPath >> (the last one being from 2019). >> >> I wonder how much of that 18% is feature detection and fallback codepaths >> <https://github.com/search?l=JavaScript&q=composedPath+event.path&type=Code> >> . >> > > Yeah, that 18% is unreasonably high... > I think it'd be good to add a non-IDL based counter that counts when both > Event::path > <https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/renderer/core/dom/events/event.cc;l=287;bpv=1;bpt=0> > and > Event::composedPath > <https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/renderer/core/dom/events/event.cc;l=293;bpv=1;bpt=0> > are > accessed. (e.g. by adding 2 booleans or a bitmask on Event indicating when > any of them was touched and triggering the counter when both have been) > Then we can assume that counts feature-detection/fallback and subtract > that counter from the 18%. > Based on past cases, I'm pretty sure most of the usage here is due to event copying, where all of the properties of event instances are copied to a new object. So I think we should basically ignore the 18% number and treat the risk as unknown. A use counter that tries to better measure the risk here would be nice, but I actually don't see how to measure it. Consider this code: function(event) { var path = event.path; if (!path) { path = event.composedPath() } doStuffWith(path); } This would never hit the composedPath() code path, and is indistinguishable to code that only relies on event.path, from the point of view of use counter instrumentation. I think that httparchive analysis here is our best tool, and in fact that *only* httparchive analysis could be enough to convince us that there's no problem here, if we fail to find any breaking cases. But even then, we should probably have a fairly long deprecation period. If we go ahead and deprecate this, what do you think the deprecation message should say about eventual removal? Best regards, Philip -- You received this message because you are subscribed to the Google Groups "blink-dev" group. To unsubscribe from this group and stop receiving emails from it, send an email to blink-dev+unsubscr...@chromium.org. To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAARdPYf2oYbdG44_jand7F5yovDZ6SjUajaQJ1xHqEmSVJ_a%2Bw%40mail.gmail.com.
