I built a pipeline by integrating my existing FME-style Python transformers as imported functions and then generating a unified Python script to execute the entire workflow. During validation, I observed inconsistencies in the generated code—specifically, certain transformer outputs are being referenced before the corresponding transformers have actually been executed.
When I run the complete pipeline as a single script, it raises errors such as NameError: name 'customTransformer18' is not defined. However, if I execute the pipeline step-by-step, each transformer runs in the correct order and produces outputs consistent with the expected FME workflow behavior.
This suggests that the issue lies in the code generation logic, particularly in how transformer dependencies and execution order are being resolved when assembling the full pipeline script.
I built a pipeline by integrating my existing FME-style Python transformers as imported functions and then generating a unified Python script to execute the entire workflow. During validation, I observed inconsistencies in the generated code—specifically, certain transformer outputs are being referenced before the corresponding transformers have actually been executed.
When I run the complete pipeline as a single script, it raises errors such as NameError: name 'customTransformer18' is not defined. However, if I execute the pipeline step-by-step, each transformer runs in the correct order and produces outputs consistent with the expected FME workflow behavior.
This suggests that the issue lies in the code generation logic, particularly in how transformer dependencies and execution order are being resolved when assembling the full pipeline script.