Flow Runner: creation of Constants delegates name creation to repr which might be memory and compute inefficient

Current behavior

In flow_runner.py, the FlowRunner’s get_flow_run_state has the following snippet

                # augment edges with upstream constants
                for key, val in self.flow.constants[task].items():
                    edge = Edge(
                        upstream_task=prefect.tasks.core.constants.Constant(val),  # <- a Constant instance is created
                        downstream_task=task,
                        key=key,
                    )
                    upstream_states[edge] = Success(
                        "Auto-generated constant value",
                        result=ConstantResult(value=val),
                    )

and under constants.py one can see that initialization of a Constant makes a call to repr(self.value); this call could be quite expensive (both in terms of compute and memory) in case the object is large and has an expensive repr method, for instance, a pandas DataFrame …

    def __init__(self, value: Any, name: str = None, **kwargs: Any):
        self.value = value

        # set the name from the value
        if name is None:
            name = repr(self.value)   # <- creates a name attribute by calling the repr
            if len(name) > 8:
                name = "Constant[{}]".format(type(self.value).__name__)

        super().__init__(name=name, **kwargs)

Proposed behavior

Since the constant’s name doesn’t seem to be too important (i.e. doesn’t show in any of the logs it seems, would it make more sense to just use a uuid generated name - which would be much more efficient than having to call on repr ? In essence the only needed code change would be the following in flow_runner.py

            # augment edges with upstream constants
            for key, val in self.flow.constants[task].items():
                edge = Edge(
                    upstream_task=prefect.tasks.core.constants.Constant(val, name=str(uuid.uuid4()),
                    downstream_task=task,
                    key=key,
                )
                upstream_states[edge] = Success(
                    "Auto-generated constant value",
                    result=ConstantResult(value=val),
                )

[cicdw]

Maybe we can use the type so that the name is still mildly informative? Names don’t need to be unique within a Flow so I think this should be safe + cheap to compute

[github-actions[bot]]

This issue was closed because it has been stale for 14 days with no activity. If this issue is important or you have more to add feel free to re-open it.