Disregard all about:<something> uris in the navigation policy.

[pvoglhofer]

We already discovered issues with about:blank requests, as these are not
regular navigation requests. These peculiarities extend to other about:
requests as well, so they should be disregarded in the decide-policy
handling too.

I'm not entirely sure this will help #80 , but it is a potential source for hanging during the 'open' call (running iframe_open.t without the changes to WebKit2.pm produces a hang)

[niner]

What happens if there is an active navigation request already and we navigate to about:blank? We can't set the active navigation request to about:blank but should we at least remove the existing (and obsolete) one?

[jscarty]

What happens if there is an active navigation request already and we navigate to about:blank? We can't set the active navigation request to about:blank but should we at least remove the existing (and obsolete) one?

Depends on how that happens. If it is set and WWW::Webkit2 activates something it will most likely get stuck inside process page load, nothing is loading so is_loading and Gtk3::events_pending will return false while active_navigation_action will always return true.

I can't think of a situation where it will be set and then navigation to an about:* page happens. I assume it should be unset when resouce-load-signal is activated.