Store configs as dictionaries

metaflow/__init__.py

@@ -103,7 +103,7 @@ class and related decorators.
 
 from .parameters import Parameter, JSONTypeClass, JSONType
 
-from .user_configs.config_parameters import Config, config_expr
+from .user_configs.config_parameters import Config, ConfigValue, config_expr
 from .user_configs.config_decorators import CustomFlowDecorator, CustomStepDecorator
 
 # data layer

metaflow/flowspec.py

@@ -24,6 +24,7 @@
 from .graph import FlowGraph
 from .unbounded_foreach import UnboundedForeachInput
 from .user_configs.config_decorators import (
+    ConfigValue,
     CustomFlowDecorator,
     CustomStepDecorator,
     MutableFlow,
@@ -310,6 +311,10 @@ def _set_constants(self, graph, kwargs, config_options):
             if isinstance(val, DelayedEvaluationParameter):
                 val = val()
             val = val.split(param.separator) if val and param.separator else val
+            if isinstance(val, ConfigValue):
+                # We store config values as dict so they are accessible with older
+                # metaflow clients. It also makes it easier to access.
+                val = val.to_dict()
             setattr(self, var, val)
             parameters_info.append({"name": var, "type": param.__class__.__name__})
 

metaflow/user_configs/config_parameters.py

@@ -338,7 +338,7 @@ def __init__(
         self._computed_value = None
 
     def load_parameter(self, v):
-        return v
+        return ConfigValue(v)
 
     def _store_value(self, v: Any) -> None:
         self._computed_value = v