The context is: #2246 (comment)
And a quote from the doc: https://docs.google.com/document/d/1EQwXL2pCuUyM1B917FUgP_8pFS3VF8F_bUfjy8IE7gM/edit?tab=t.vmaefhinvkl5
To prevent a complicated dependency graph, any plugin that implements the PrepareData hook point, may not implement any other hook points that would relate to pool-level data production. This is to prevent uncertainty when building a dependency graph.
We must ensure that the DAG validation we generate is informative & robust so that users can rely on the data system thoroughly.
The context is: #2246 (comment)
And a quote from the doc: https://docs.google.com/document/d/1EQwXL2pCuUyM1B917FUgP_8pFS3VF8F_bUfjy8IE7gM/edit?tab=t.vmaefhinvkl5
We must ensure that the DAG validation we generate is informative & robust so that users can rely on the data system thoroughly.