Once we know the sheet's principal (bug 221428) we should propagate it to URI CSSValues.

Comment on attachment 263439 [details] [diff] [review] Like so DOMCSSDeclarationImpl::GetCSSParsingEnvironment shouldn't rely on the caller null-initializing the out param. (And maybe the other GetCSSParsingEnvironment should null the out param in failure cases as well, for symmetry.) The XXXbz comment in nsGenericHTMLElement and the comment in nsXULElement about doc vs. node scare me a little. Are those going to be an issue with XBL2? Should the latter have an "XXX"? Should followup bugs be filed?

> DOMCSSDeclarationImpl::GetCSSParsingEnvironment shouldn't rely Will do. Good catch. > Are those going to be an issue with XBL2? I don't know yet. Right now, |node->NodePrincipal()| and |node->GetOwnerDoc()->NodePrincipal()| are equivalent, unless the node's owner document is gone already (in which case the former is what it's principal used to be and the latter crashes). I suspect we'll keep this setup with XBL2, and since the owner document will be the XBL document, things will work fine. That said, once we start implementing XBL2 we'll need to audit existing NodePrincipal() callers no matter what, to catch ones using the GetCurrentDoc().... The comments in this patch are largely a guide for then.

Checked in.