Something done on the screen that updates what the whole screen shows: the
run answers the screen's fresh content, the engine rebuilds the screen's
handle over it and answers that handle, and the caller continues from it —
its state is the content's, so the next action, edge or checkpoint continues
from what the update changed (a keystroke that leaves the search field with
its typed results and session).
The handle the action was called on keeps the snapshot the visit loaded and
is no longer the screen on top: like a page a feed answers, the answer is the
handle to continue with.
The screen's item lists are shown over the answer, so their enters run —
a list an action refreshes alone is answered with item handles
(itemHandlesOf) instead. A screen that keeps a frame position beyond what
entering showed (a feed's depth) does not declare it, because the frame's
position still belongs to the handle it was entered with.
Something done on the screen that updates what the whole screen shows: the run answers the screen's fresh content, the engine rebuilds the screen's handle over it and answers that handle, and the caller continues from it — its state is the content's, so the next action, edge or checkpoint continues from what the update changed (a keystroke that leaves the search field with its typed results and session).
The handle the action was called on keeps the snapshot the visit loaded and is no longer the screen on top: like a page a feed answers, the answer is the handle to continue with.
The screen's item lists are shown over the answer, so their
enters run — a list an action refreshes alone is answered with item handles (itemHandlesOf) instead. A screen that keeps a frame position beyond what entering showed (a feed's depth) does not declare it, because the frame's position still belongs to the handle it was entered with.