Fixing failing silently

[mpan322]

For Issue: #20

Possible Failure Cases:

• Add node
  • if node exists with the same name it should fail. ✓
  • if the node's name is not a valid node identifier it should fail immediately X (at the moment there is only very basic checking done when stringifying) . I am working on code to check based on the graphviz specification.
• Add edge
  • if either of the nodes do not exist it automatically creates the node. I think this is fine as it is much more convienient. ✓
  • it should also allow for node failure cases above based on their names ✓
• Remove node
  • if the node does not exist it should fail ✓
• Remove edge
  • if no such edge exists it should fail ✓
• Add subgraph / adding a context
  • if there is already a subgraph / context with the same name within the same parent it should fail ✓
    • QUESTION: At the moment this check is only done within the parent i.e. if g has subgraphs a and b it is valid for both a and b to have children labelled c.
    • I am partial to removing this as it will likely make the subgraph tree searching operations difficult to use properly or not make sense in their current version. Let me know if you think there is a good reason to keep this behaviour.
• Add attribute
  • if no known attribute exists there is a warning (discussed this previously) ✓
• Remove attribute
  • removing an attribute which is not set causes in error ✓
• Set attribute (which is already set)
  • overwrites but does not change warn or error (I find this behaviour fine - let me know if you disagree) ✓
  • CHANGE: made it so adding an attribute to a graph now follows the above behaviour (before it was not removed)

Commits

• Added errors and updated tests for the following cases: - removing edges between nodes which do not exist. - removing attributes which have not been set.
• 1. Added quotations arround node names in errors. 2. When removing edges between nodes in a graph if no edges exist and error is outputted.
• - Added function to check if a label is valid. - Reduced some code duplication in other spots TODO: add label validity checking to all functions where labels may be added.
• added planning
• added attribute removal when setting an existing attribute on graphs, added more tests and went through operations which could fail and reasonably should error