Remove initEvent quirk
Created attachment 297558 [details] Patch
Created attachment 297563 [details] Patch
http://trac.webkit.org/changeset/210045
Is there a radar tracking this?
rdar://problem/29420268
Comment on attachment 297563 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=297563&action=review > Source/WebCore/ChangeLog:9 > + Many web compatibility issues have been found, which leads us to believe that many more exist. I disagree with this comment, there were no Web compatibility issues found. It merely broke several iOS apps that are WebKit-specific and relied on our bug. Because of this, I believe it is going to be difficult the get the specification changed. It is also going to be difficult to get the specification changed because every other browser out there matches the specification. I don't mind trying to get the spec changed but until it does, I believe the way forward for us is to use the quirk for iOS apps, not *regress* our standards compliance and interoperability with other browsers.
Comment on attachment 297563 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=297563&action=review > Source/WebCore/dom/Event.idl:69 > + void initEvent(optional DOMString type = "undefined", optional boolean bubbles = false, optional boolean cancelable = false); Also I doubt there is any app calling initEvent without the first parameter.
I agree with Chris’s comments.
(In reply to comment #8) > I agree with Chris’s comments. Alex pointed out that Chrome does not match the spec either so we are discussing this upstream at https://github.com/whatwg/dom/issues/387. I will follow up once we reach agreement at spec-level.
I didn't find any content calling initEvent without the first parameter, they were just missing the second and third. This just reverted completely to our old behavior. I think we should definitely become spec compliant, but this doesn't seem to cause interoperability issues; I haven't seen any content that relies on a JavaScript exception when there are fewer than three parameters.
Sounds like a good approach.