More TODO: graphql

[iamsmkr]

graphql

• [x] wait_for_server needs to be internalized
• [x] wait to be internalised into stop itself in python
• [x] change graphs api to return graphqls instead of vec of graphs
• [x] add precedence to config load and impl default configs
• [x] add tests graphql python
• [x] impl and add tests for client/server apis https://github.com/Pometry/Raphtory/pull/1656#issuecomment-2192094526
• [x] Move the log_level to the server object so all config is there
• [x] Remove timeout from run and set a sensible default (2 mins or something)
• [x] Change GQLGraphs.names to name and make this check the names of all graphs within its working_dir
  • [x] ensure any questions to GQLGraphs brings in all graphs fromthe working_dir
  • [x] For the second remove any other functions on GQLGraphs as I think they could be a little dangerous
• [x] for the RaphtoryServer.run_with_auth set it to return a result and replace all of the expects with Err
• [x] Remove Graphs hashmap/paths from the RaphtoryServer so all uploads are done via the client
• [ ] Sort the different ways you have give graphs to the server, these should be:
  • [x] provide a hashmap where the keys are the filenames - these should all be set via the multipart upload as the binary string seems very slow
  • [x] provide a collection of graph files (parsed via glob (https://docs.rs/glob/latest/glob/)) for a path ON the server.
  • [x] Provide a collection of graph files on the local machine which can be uploaded via multipart upload
  • [ ] All of the above functions should take a namespace argument which is the path that the graphs should be saved in within the working_dir
  • [x] All of the above should have the overwrite argument set to false by default where we first check if all of the names are available within the namespace and if not an error is returned
  • [ ] Can we have a basic 'write_access' conf which enables or disables the three above functions. The two flags should be client_upload and server_read

Testing to do later

• [ ] improve test coverage including both storage and non-storage paths
• [ ] test if server started on localhost is accessible via container
• [ ] impl graphql benchmarking
• [ ] add multi-client test

[miratepuffin]

• We dont seem to make use of upload_graph from the client - which should be a good bit faster than send_graph
• The load_graphs_from_path from the server and client seems to crash:

[iamsmkr]

• The load_graphs_from_path from the server and client seems to crash:

Seems to be working fine in branch todos/graphql

[miratepuffin]

server = RaphtoryServer(working_dir="/tmp/graphs/",port=1736)
server.start()
c = server.get_client()
c.send_graph(g,graph_name="blah",namespace="ben/tests/",overwrite=False)
c.upload_graph("/user/ben/Documents/g",graph_name="blah2",,namespace="ben/tests/",overwrite=False)
c.load_graph("/usr/ubuntu/g",graph_name="blah3",,namespace="ben/tests/",overwrite=False)

From discussions on simplifying the upload of graphs