Description
It is high time that gremlin-python comes packaged with a real driver. After watching the community discussion, it seems that the way to go will be to use the concurrent.futures module with multithreading to provide asynchronous I/O. While the default underlying websocket client library will remain Tornado due to Python 2/3 compatibility issues, this should be decoupled from the rest of the client and easy to replace.
With this is mind, I created a baseline client implementation with this commit in a topic branch python_driver. Some things to note:
- All I/O is performed using the concurrent.futures module, which provides a standard 2/3 compatible future interface.
- The implementation currently does not include the concept of a cluster, instead it assumes a single host.
- The transport interface makes it easy to plug in client libraries by defining a simple wrapper.
- Because this is an example, I didn't fix all the tests to work with the new client implementation. Instead I just added a few demo tests. If we decide to move forward with this I will update the original tests.
The resulting API looks like this for a simple client:
client = Client('ws://localhost:8182/gremlin', 'g')
g = Graph().traversal()
t = g.V()
future_result_set = client.submitAsync(t.bytecode)
result_set = future_result_set.result()
results = result_set.all().result()
client.close()
Using the DriverRemoteConnection:
conn = DriverRemoteConnection('ws://localhost:8182/gremlin', 'g')
g = Graph().traversal().withRemote(conn)
t = g.V()
results = t.toList()
conn.close()
If you have a minute to check out the new driver code that would be great, I welcome feedback and suggestions. If we decide to move forward like this, I will proceed to finish the driver implementation.
Attachments
Issue Links
- links to