jgraham

The general tl;dr of the telecon discussion is that the element reference cache only needs to live as long as the page (from the point of view of modern browser...

Based on my understanding of where we're at with this issue, I've updated the issue summary. If we make `list of known elements` per Window rather than per browsing context,...

geckodriver supports this, but it's a new endpoint rather than a parameter (`…/screenshot/full`). Since the existing endpoint is a `GET` and we don't have other cases that use parameters on...

Tests in https://phabricator.services.mozilla.com/D152645

> I'm not sure whether geckodriver wants control over that directory, to read/write things there too? Yes, geckodriver needs to be able to set prefs in the profile, so they...

You still need to specify where you want to put the profile, because the snap doesn't share `/tmp` with the host. We could perhaps pick a different default in this...

https://bugzilla.mozilla.org/show_bug.cgi?id=1678347

Uh, the spec seems to assume all element properties have a trivial serialization. That seems wrong. Pretty sure anything that can return an arbitary JS object should be calling https://w3c.github.io/webdriver/#dfn-json-clone...

We should certainly consider bfcache.

`"readOnly"` isn't the name of the attribute, so I think our behaviour is correct per spec. If this isn't the required behaviour here we should have a discussion in a...