Open yamalight opened 4 weeks ago
Did some thinking and I'm so far not convinced this is 100% required. Would freeze this until there's a specific use case that cannot be done without it.
Engine should be split into two components - pipeline assembler and pipeline executor. Executor should always be the same and work the same way regardless of assembler behavior. Assembler should be replaceable and should return a pipeline executor can run. I think this approach will allow for some cool things.
Using RxJS as core would like work very well.
Implement advanced step execution flows:
Use cases: