To ensure `handle_peer_queue` and `handle_dead_peer_queue` are indeed
run before the tests finish. Previously, we get errors when performing
`iter_dag` after cancellation. This is because `handle_peer_queue` or
`handle_dead_peer_queue` is not actually run before the Service is
cancelled.
- Dedup `perform_test_from_obj` and the test cases used in both
`test_floodsub` and `test_gossipsub_backward_compatibility.py`.
Therefore, they are put in the standalone file
`tests/pubsub/floodsub_integration_test_settings.py`. The functions
and testcases are imported from there then.
- IMO still need a refactor on the tests. There are still some duplicate
code.
* Add handle_rpc call to pubsub
* Scaffold gossipsub functions
* Add timer
* Implement most of mesh construction
* Implement emit and handle
* Implement fanout heartbeat
* Refactor emit
* some gossipsub cleanup and test
* minor lint stuff, more to come
* Implement publish
* Fix comment
* Modify pubsub/gossipsub so that floodsub tests pass using gossipsub router
* Add floodsub tests to gossipsub
* Handle case where select_from_minus, num_to_select > size(pool-minus)
* Add topic membership
* Implement handle ihave
* Implement most of iwant
* Add mcache.add and comments
* Refactor handle_ihave
* Implement stream write in handle_iwant
* Implement gossip heartbeat
* unresolved vars
* initial mcache code
* documenting mcache
* writing test/debugging mcache
* finished mcache test and debugged
* Make gossipsub backward compatibility its own file
* remove mcache prints
* DEBUGGING
* Add sender_peer_id to handle_rpc to get gossip test passing
* Modify gossipsub to make fanout work
* fanout maintenance test
* debugging gsub GOSSIP
* DEBUGGING
* debugged sender seen cachce
* adding lru, removing prints
* pylint cleanup
* Fix github comments in PR
* minor floodsub possible bugfix
