Problem found in graphs.is_planar() when the graph has no edges.

[edward.scheinerman]

The is_planar method for graphs includes a set_embedding argument that saves the combinatorial embedding of the graph (if it's planar). In this example, it works properly:

sage: g = graphs.WheelGraph?(6) sage: g.is_planar(set_embedding=True) True sage: g.get_embedding() {0: [1, 5, 4, 3, 2], 1: [2, 5, 0], 2: [0, 3, 1], 3: [0, 4, 2], 4: [0, 5, 3], 5: [0, 1, 4]}

However, is we apply this method to a graph with no edges, a problem arises:

sage: g = graphs.EmptyGraph?() sage: g.add_vertices(range(6)) sage: g.is_planar(set_embedding=True)

---

AttributeError? Traceback (most recent call last)

/home/ers/<ipython console> in <module>()

/usr/public/share/sage/local/lib/python2.6/site-packages/sage/graphs/generic_graph.pyc in is_planar(self, on_embedding, kuratowski, set_embedding, set_pos)

2438 self._pos = G._pos 2439 if set_embedding:

-> 2440 self._embedding = G._embedding

2441 return planar 2442

AttributeError?: 'Graph' object has no attribute '_embedding'

Once we add an edge to this graph, all is well:

sage: g.add_edge(0,1) sage: sage: g.is_planar(set_embedding=True) True sage: g.get_embedding() {0: [1], 1: [0], 2: [], 3: [], 4: [], 5: []}

[brunellus]

The patch from #12134 should solve this problem.