Comment 36 for bug 1022741

Revision history for this message
In , Imelven-s (imelven-s) wrote :

Created attachment 585561
WIP - reworked prototype

- nsIDocument, nsIDocShell, nsIDOMHTMLIFrameElement IIDs bumped

- added nsHTMLIFrameElement::FromContent(nsIContent*)

- added PRUint32 nsHTMLIFrameElement::GetSandboxFlags()

- to better match the HTML5 spec, break the flags up more, reverse their polarity and rename them.

- change sandbox attribute parsing to use nsCharSeparatedTokenizerTemplate<nsContentUtils::IsHTMLWhitespace>

- add NULL check to fix crash with chrome://mochitests/content/browser/toolkit/mozapps/extensions/test/xpinstall/browser_navigateaway2.js

- lots of other misc. cleanup from feedback from ms2ger and khuey

next up:

doing some more debugging of the MS test suite tests that I think should be passing but are not - and then moving on to document.cookie and blocking it appropriately.

The big outstanding question at the moment : do I want to make the tests for this manual (as the Microsoft test suite is) or attempt to make automated tests ? Manual tests have the problem that they don't always get run on every checkin - automated tests would need (I think) some sort of event that happens when something is blocked (CSP does this with an observable notification, for example) - the con here is this doesn't test the thing attempting to be blocked actually is blocked, just that the event was fired.