… suite

Uses Gherkin to write test specifications that will be implemented by the various GLVs. Provided a basic implementation for gremlin-python.

This matches he pattern of the java test suite.

Developed methods for vertices/maps and a way to assert unordered results.

… vertex object

This might be the pattern to use across the board. We'll see if there is a better idea floating about though so may not be final.

Included infrastructure for validating maps and refactored other related code.

Included a way to specify element identifiers in feature files.

…as features

…iles

Tests should be faster now that remotes and data are cached for the toy graphs.

Needed to add support for cleaning up traversal strings to more properly handle reserved python words.

Sorta made the decision to be explicit in how traversals are defined in .feature files. Shying away from using shorthand static imports which will make it more explicit for those trying to implement the tests.

Included option for next() in the test language.

Used "defined as" instead of "is" since the line reads more clearly with the former.

Graph init language provides a way to add data to an empty graph as part of a scenario.

… tests

Needed to account for floats vs longs with d[x]

Had to add a way to assert a results that matched any in a set as limit() type steps might force different results depending on the graph being tested.

Reworked test language for asserting counts so that it works better as a standalone assertion.

Tables now return a list of maps that required a header so all feature files needed header.

Minor refactoring to test language around result counts

No point to enabling at this time since all tests aren't really targetted for migration with this initial body of work. It will be turned on later when all test are intended for migration.

Added fixes to python test logic for the reserved global keyword and fixed ordering test logic a bit.

This is not complete, but it is enough for now.

Included Cardinality in imports for test logic

Tests were failing as a result of the change to using the mix server configuration that had all the graphs. On the way to dealing with that, I noticed the driver didn't seem to have aliasing capability which prevented it from choosing the correct graph traversal source on the server. For some reason, asserting longs on ids seemed to be a problem as well after this change and I'm pretty sure it has something to do with the configuration of the TinkerGraph in this new mixed mode configuration and not something in serialization.

Ensures that a RemoteConnection in .NET is properly initialized.