eval() function does not try all declared types

[BClappe]

Before submitting the issue

• I have checked for Compatibility issues
• I have searched among the existing issues
• I am using a Python virtual environment

Description of the bug

eval() function seems not to try all declared types for a pin.

Steps To Reproduce

Let's use the export_symbolic_workflow operator:

import ansys.dpf.core as dpf
import ansys.dpf.core.operators as ops

serializer = ops.serialization.export_symbolic_workflow(workflow=dpf.Workflow())
content = serializer.eval()  # Crashes with: ansys.dpf.gate.errors.DPFServerException: The requested Data format"data_sources" is not valid for this pin

For the only output pin, the declared types are: "data_sources | string". In this case, we want a string (denoted by the fact we didn't give any file path on pin 2), but apparently only "data_sources" has been tested, and crashed.

Which Operating System causes the issue?

Windows, Linux

Which DPF/Ansys version are you using?

Ansys 2027 R1

Which Python version causes the issue?

3.13

Installed packages

ansys-dpf-core==0.14.2

[BClappe]

Ok, apparently the double type is not really a double type, but rather a conditional type.
This bug seems to be a confusion on the meaning of the specifications and the behaviour of the operators. @PProfizi
When a filepath is specified, only "data_sources" is available, and when no filepath is specified, only "string" is available.

See:

import ansys.dpf.core as dpf
import ansys.dpf.core.operators as ops

serializer = ops.serialization.export_symbolic_workflow(workflow=dpf.Workflow())

content = serializer.get_output(0, dpf.types.string)  # Works
content = serializer.eval()  # Crashes
content = serializer.get_output(0, dpf.types.data_sources)  # Crashes

And:

import ansys.dpf.core as dpf
import ansys.dpf.core.operators as ops

serializer = ops.serialization.export_symbolic_workflow(workflow=dpf.Workflow(), path="something")

datasource = serializer.get_output(0, dpf.types.data_sources)  # Works
datasource = serializer.eval()  # Works: type data_sources
datasource = serializer.get_output(0, dpf.types.string)  # Crashes