226725 2021-06-07 04:47:08 -0700 MouseEvent.constructor misses checking if the param value is non-infinite or not 2021-06-16 10:59:38 -0700 1 1 1 Unclassified WebKit WebKit Misc. WebKit Local Build Unspecified Linux RESOLVED WONTFIX InRadar P2 Normal --- 1 zyscoder webkit-unassigned cdumez graouts sam webkit-bug-importer oldest_to_newest 1767409 0 zyscoder 2021-06-07 04:47:08 -0700 Steps to reproduce: (1) Open a tab and navigate to any URL; (2) Run the following code in the Console of Devtools: ``` new MouseEvent('',{screenX: NaN}); ``` (3) This code would be evaluated successfully without any exceptions thrown. Actual results: This code would be evaluated successfully without any exceptions thrown. Expected results: In Chome, this code would throw an exception: `Uncaught TypeError: Failed to construct 'MouseEvent': The provided double value is non-finite.`. It seems reasonable to throw such an exception since the type of the screenX param is double. 1767639 1 zyscoder 2021-06-07 19:35:02 -0700 As https://docs.w3cub.com/dom/mouseevent/screenx says, screenX is `A double floating point value. Early versions of the spec defined this as an integer referring to the number of pixels`. 1767847 2 sam 2021-06-08 11:57:05 -0700 Looks like the latest version of UIEvents still has this as a long: https://w3c.github.io/uievents/#idl-mouseeventinit. 1768035 3 cdumez 2021-06-08 20:09:33 -0700 (In reply to Sam Weinig from comment #2) > Looks like the latest version of UIEvents still has this as a long: > > https://w3c.github.io/uievents/#idl-mouseeventinit. Indeed, we are matching the specification and Firefox. Maybe this a bug should be filed against Chrome? They are the only ones not aligned here. 1769353 4 webkit-bug-importer 2021-06-14 04:48:20 -0700 <rdar://problem/79282818> 1770276 5 cdumez 2021-06-16 10:59:38 -0700 We are aligned with Firefox and the specification here. I don't think we should make this change for now.