Error creating an empty multi-edge (di)graph

[sagetrac-mhs]

In SAGE 4.5.1 if you try to create an empty graph or digraph using the multi-edge keyword the (di)graph created is a non-multi (di)graph.

Examples of the strange behaviour:

sage: G=Graph(multiple_edges=True)
sage: G
Graph on 0 vertices

sage: G=Graph([],multiple_edges=True)
sage: G
Graph on 0 vertices

sage: H=DiGraph(multiple_edges=True)
sage: H
Digraph on 0 vertices

sage: H=DiGraph([],multiple_edges=True)
sage: H
Digraph on 0 vertices

Nevertheless it works fine, i.e. produces a multi-graph/multi-digraph if one uses an empty dictionary like in the following:

sage: G=Graph({},multiple_edges=True)
sage: G
Multi-graph on 0 vertices

sage: H=DiGraph({},multiple_edges=True)
sage: H
Multi-digraph on 0 vertices

After posting on sage-develop Strange behaviour of add_edges http://groups.google.com/group/sage-devel/browse_thread/thread/69773eae8ba3cefc I was asked to open a new ticket...

CC: @sagetrac-mvngu @sagetrac-brunellus

Component: graph theory

Keywords: creating empty Graph, DiGraph, multiple_edges=True

Author: Lukáš Lánský

Reviewer: Nathann Cohen

Merged: sage-5.0.beta3

Issue created by migration from https://trac.sagemath.org/ticket/10441

[sagetrac-mvngu]

Changed upstream from Not yet reported upstream; Will do shortly. to none

[sagetrac-mvngu]

Changed author from MHS to none

[sagetrac-brunellus]

comment:4

I think that having **kwds in the Graph constructor is very confusing, because any misspelling of an argument pass through without notice. Do you see some serious reason why is this there? I'm looking into the code and can't see anything.

[sagetrac-brunellus]

comment:5

I started #12318 for the mentioned problem.

Fix for this ticket is a rather simple one, I think. Prove me wrong! :-)

[nathanncohen]

comment:6

Hellooooooooooooo !!

Well, the code building the graph is a bit messy but indeed what you wrote 1) prevents the bug 2) is what we should do until somebody comes with another weird situation :-)

Would it be possible to move your "# not a multi [...]" comments to the line just above the test, though ? That's where we put flags like "optional" or "not tested" for the automatic tests, and it's best to be sure these comments do not interfere ^^;

Nathann

[sagetrac-brunellus]

comment:8

Attachment: trac_10441_multigraph.patch.gz

Thanks! Is it better now?

[nathanncohen]

comment:9

Hellooooooo !!!

I thought you would put the comments just before the "::" but there's no problem with way either... Good to go ! :-)

Nathann

[nathanncohen]

Reviewer: Nathann Cohen

[nathanncohen]

Author: Lukáš Lánský

[jdemeyer]

Merged: sage-5.0.beta3