I think the current way of generating it via a fixed jinja template file and hard coded lines in the BootscriptBuilder class is too restrictive.

[nanaones]

Is your feature request related to a problem? Please describe.

I've been working with Elyra lately and have exported and re-edited the pipeline as a Python 'DAG' file. As data platform providers, we felt this was a stumbling block for our users. Therefore, we are considering adding a 'catalog' to the task to solve this problem. However, when working by adding 'Catalog', there is a problem that the execution result through the 'Execute as pipeline' button and the execution result through pipeline execution in the 'Pipeline editor' are different. We think this is a very poor user experience.

Describe the solution you'd like So, I would like to propose a way to find out the location of the jinja Template file through environment variables and change it so that it can be used, and receive and write arguments of the BootscriptBuilder class from the outside.

Describe alternatives you've considered I think this list can be considered outside the elyra library. https://github.com/elyra-ai/elyra/blob/master/elyra/airflow/operator.py#L97

And this line too.

https://github.com/elyra-ai/elyra/blob/main/elyra/pipeline/airflow/processor_airflow.py#L501

TEMPLATE_FOLDER_PATH = os.environ.get("TEMPLATE_FOLDER_PATH", False)
TEMPLATE_FILE = os.environ.get("TEMPLATE_FILE", "python_dsl_template.jinja2")

if TEMPLATE_FOLDER_PATH:
    loader = FileSystemLoader(searchpath=TEMPLATE_FOLDER_PATH)
else:
    # Load Kubeflow Pipelines Python DSL template
    loader = PackageLoader("elyra", "templates/kubeflow/v1")