WebCore will need to know if a page is automated to expose extra WebDriver API on the navigator object. WebKit2 also needs to know for Safari's injected bundle code.
<rdar://problem/24965784>
Created attachment 272793 [details] Patch
Comment on attachment 272793 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=272793&action=review r=me > Source/WebKit2/UIProcess/API/C/WKPagePrivate.h:88 > +WK_EXPORT bool WKPageGetIsControlledByAutomation(WKPageRef page); > +WK_EXPORT void WKPageSetControlledByAutomation(WKPageRef page, bool controlled); It seems a little roundabout to require a WK2 API for a client to decide the page is for automation. I realize that is the case right now, but perhaps we should file a bug / have a comment somewhere that we should move to having automation pages in their own process pool, so it is guaranteed / known in WebKit what pages are automation without requiring client involvement.
Comment on attachment 272793 [details] Patch Clearing flags on attachment: 272793 Committed r197564: <http://trac.webkit.org/changeset/197564>
All reviewed patches have been landed. Closing bug.
Comment on attachment 272793 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=272793&action=review >> Source/WebKit2/UIProcess/API/C/WKPagePrivate.h:88 >> +WK_EXPORT void WKPageSetControlledByAutomation(WKPageRef page, bool controlled); > > It seems a little roundabout to require a WK2 API for a client to decide the page is for automation. I realize that is the case right now, but perhaps we should file a bug / have a comment somewhere that we should move to having automation pages in their own process pool, so it is guaranteed / known in WebKit what pages are automation without requiring client involvement. While in general I agree that it would be nice to confine automation pages to their own process pool, this may not be feasible for all ports. I'l file a bug internally to get feedback. <rdar://problem/24996361>