A blazingly fast drop-in replacement for the Mastodon streaming API server
Go to file
Daniel Sockwell 30453c1fb7 Update README with notes on code architecture 2019-04-28 17:44:59 -04:00
src Add ability for multiple clients to connect to the same pub/sub connection 2019-04-28 17:28:57 -04:00
.gitignore Initial project files 2019-02-11 09:45:14 +01:00
Cargo.lock Add ability for multiple clients to connect to the same pub/sub connection 2019-04-28 17:28:57 -04:00
Cargo.toml Add ability for multiple clients to connect to the same pub/sub connection 2019-04-28 17:28:57 -04:00
LICENSE Initial commit 2019-02-08 10:35:26 +01:00
README.md Update README with notes on code architecture 2019-04-28 17:44:59 -04:00

README.md

RageQuit

A blazingly fast drop-in replacement for the Mastodon streaming api server

Notes on data flow

The current structure of the app is as follows:

Client Request --> Warp Warp filters for valid requests and parses request data. Based on that data, it repeatedly polls the StreamManager

Warp --> StreamManager The StreamManager consults a hash table to see if there is a currently open PubSub channel. If there is, it uses that channel; if not, it creates a new channel using the methods in pubsub.rs. Either way, it ends up with a Receiver to poll. The StreamManager polls the Receiver, providing info about which StreamManager it is that is doing the polling. The stream manager is also responsible for monitoring the hash table to see if it contains any Receivers that no longer have active clients; if it does, the StreamManager removes them from the hash table (which causes them to be dropped from memory and causes the PubSub connection to be closed).

StreamManger --> Receiver The Receiver receives data from Redis and stores it in a series of queues (one for each StreamManager). When polled by the StreamManager, it sends back the messages relevant to that StreamManager and removes them from the queue.