On Fri Oct 21 18:59:31 2022 +0000, Gabriel Ivăncescu wrote:
For stubs, I generally avoided those that don't have some respective gecko interface, because I was a bit unsure how their implementation would actually look like. Some of the others aren't forwarded because they used some fields from the former `basedoc`, which are different on HTMLDocumentObj compared to node (most commonly "window"), and I didn't want to change the existing behavior at all, but keep it no-op. As for the "other way around", seems it's just get_referrer; since this is likely tied to the window, I wanted to avoid forwarding it to the node (same reason as above). As I understand, the doc object is always available, but the node might not always be. It should be fine this way? Which stubs did you have in mind to forward, only the IHTMLDocument* ones, or also for the other interfaces?
I mostly meant DOM APIs, those are all IHTML* interfaces. And yes, for functions returning window object it may be a bit more tricky. This will need more testing (and possibly changes to window object), so we may keep it for now. But that doesn't really apply to get_referrer(). Even if window representation may differ, top document node and document object still refer to the same actual window with the same navigation history, so it should be fine to forward to node.