When windows are triggered, there a bunch of conditions that might halt the immediate firing of algorithms.
They are:
- Waiting for results to complete so they can be factored in to lookback dependencies
- Rate limiting specified by the processors (new feature)
- Others?
We are currently writing all window and algorithm results ahead of time, and then just updating with status/results. What needs to be implemented on top of this is a work queue that allows that preregistering to happen, but then first checks the go/nogo status, and then waits for conditions to be fufilled before continuing.
This issue captures these required changes.
When windows are triggered, there a bunch of conditions that might halt the immediate firing of algorithms.
They are:
We are currently writing all window and algorithm results ahead of time, and then just updating with status/results. What needs to be implemented on top of this is a work queue that allows that preregistering to happen, but then first checks the go/nogo status, and then waits for conditions to be fufilled before continuing.
This issue captures these required changes.