Remove Blink>DOM>Events

dtapuska@, do you know anything about this event?

We have UMA for textinput event:

https://www.chromestatus.com/metrics/feature/popularity#TextInputFired
0.0872%

And more:
- TextInputEventOnInputObsolete  0.1844%
- TextInputEventOnInput 0.0245%
- TextInputEventOnTextAreaObsolete 0.0213%
- TextInputEventOnTextArea 0.0065%
- TextInputEventOnContentEditableObsolete 0.0095%
- TextInputEventOnContentEditable 0.0697%
- TextInputEventOnNotNode <=0.0001%

It seems >= 0.003% :-<

There's also https://www.chromestatus.com/metrics/feature/timeline/popularity/831 which is how often the data attribute is accessed, which means that the event listener is actually doing something. That's at ~0.08%.

We don't have a fixed threshold for removal, but usage is certainly high enough that we would want to understand why, and what the breakage would be.

It looks like WebKit and EdgeHTML also have a TextEvent interface, although I haven't checked when events are fired. A good next step here would be to write some basic tests for when these events are fired, and see how interoperable it already is between the 3 engines that do support it.

Is this event supposed to be made obsolete by the new beforeinput and input events?

FYI, removing textinput is already mentioned in issue 382814.

The textinput event is used in react:
https://github.com/facebook/react/blob/e96dc140599363029bd05565d58bcd4a432db370/packages/react-dom/src/events/BeforeInputEventPlugin.js#L43

Note that they have a fallback in the event that textinput is not available:
https://github.com/facebook/react/blob/e96dc140599363029bd05565d58bcd4a432db370/packages/react-dom/src/events/BeforeInputEventPlugin.js#L404

It's possible that react is the reason for these high usage numbers, access of the data attribute (when textinput is detected as available) is here:
https://github.com/facebook/react/blob/e96dc140599363029bd05565d58bcd4a432db370/packages/react-dom/src/events/BeforeInputEventPlugin.js#L300

So, we could test if there would be any breakage for react if we were to remove textinput. Would it make sense to create a use counter which measures non-react usage (caveat: I don't know how we would do this - maybe the script name)?

Also, we could modify react to prefer using the beforeinput event which has been implemented ( issue 585875 ).