This PR is about the wip concerning the refacto of the graph building overhaul. In an effort to make the make_subgraph method private, I also changed 2-3 things in the way the graph is built. The thing I dont' like is that every call we have to re add/remove debug nodes, + we rely on the deepcopy too much and may have overlooked edge cases, but this will be the work of another PR. This PR adds boilerplate code to achieve the aforementioned goal, which highlights a problem in the design of DAGExecution, its logic being not far away from the dag yet we're obliged to re-do some operations, this will need to be refactored in the future.
Description
This PR is about the wip concerning the refacto of the graph building overhaul. In an effort to make the
make_subgraphmethod private, I also changed 2-3 things in the way the graph is built. The thing I dont' like is that every call we have to re add/remove debug nodes, + we rely on the deepcopy too much and may have overlooked edge cases, but this will be the work of another PR. This PR adds boilerplate code to achieve the aforementioned goal, which highlights a problem in the design of DAGExecution, its logic being not far away from the dag yet we're obliged to re-do some operations, this will need to be refactored in the future.