@acsor

I know it's three years later, but for the record (meaning, more for anyone else who comes across this issue in the future): Yes, Pydot does guarantee that nodes — and all other entities — will be output in the order they were inserted into a graph with add_node(), add_edge(), add_subgraph(), etc.

A monotonically-increasing sequence number is attached to each element at the time of insertion. When the graph source is generated, all elements are output in that sequence order. Deleted nodes are simply left as gaps in the sequence. If the same element is added twice, or deleted and re-added, the second addition will be placed at the end of the sequence.

(The reason those guarantees are made is, the ordering is critical for the creation of proper graphviz source code. For example, default attribute statements — node, edge, or graph nodes — will only affect the elements that come after them in the source.)

So, there's a big difference between this:

graph G {
    node [shape=box];
    edge [color=red];
    A -- B;
    edge [color=green];
    B -- C;
}

and, say, this:

graph G {
    A -- B;
    B -- C;
    edge [color=green];
    node [shape=box];
    edge [color=red];
}

@lkk7: (And with that, I think this can probably be closed as answered?)