Changelog
release/5.0.0
Breaking Changes
-
Removed
parallelargument fromnode.run(): Usemax_workersparameter instead to control parallel execution. Setmax_workers=1for sequential execution ormax_workers=Nfor parallel execution with N workers. You can also setFLYPIPE_NODE_RUN_MAX_WORKERSenvironment variable to configure this globally (defaults to1).- RemovedFLYPIPE_DEFAULT_RUN_MODEconfig: This config previously defined parallel or sequential execution mode. Now useFLYPIPE_NODE_RUN_MAX_WORKERSinstead to control parallelism. Set it to1for sequential execution orNfor parallel execution with N workers. This can be set globally via environment variable or passed directly tonode.run()as themax_workersparameter.- UpdatedCacheclass method signatures: Theread()andwrite()methods now include optional CDC (Change Data Capture) parameters: -read(from_node=None, to_node=None, *args, **kwargs): Addedfrom_nodeandto_nodeparameters for CDC filtering -write(*args, df, upstream_nodes=None, to_node=None, datetime_started_transformation=None, **kwargs): Addeddf,upstream_nodes,to_node, anddatetime_started_transformationparameters for CDC metadata tracking - All custom cache implementations must update their method signatures to include these parameters, even if CDC functionality is not used (parameters can be ignored)- Addedcreate_cdc_table()method toCacheclass: A new optional method for caches with CDC support. The base implementation is a no-op, so existing cache implementations will continue to work, but cache classes that support CDC should override this method to create their CDC metadata tables.Commits