There are minimal dependencies on target_ at this point, and it looks like we can (almost) remove it.
The biggest issue is what to do with the unique name when building the HistoryEntry from PageState. Fortunately, it looks like all the HistoryNodes are owned by a unique_ptr, so we can probably just store it externally: this should allow layout test helpers to continue referring to it, while dropping the need to plumb it in and out of Blink.
Comment 1 by dcheng@chromium.org
, Sep 30 2017