build_component_spec_for_group raises TypeError when using Input[List[Dataset]] within dsl.If

[heavenmarshal]

Hi, I intended to implement a ETL pipeline that has multiple data sources and one data processor. The pipeline updates the data watermarks based on the returned state of the data processor. But the compiler raises a TypeError. The minimum reproducible example is in the code section below

# test_pipeline.py
from typing import List

from kfp import compiler, dsl
from kfp.dsl import Dataset, Input, Output


@dsl.component(base_image="python:3.9")
def source_component(num: int, data: Output[Dataset], watermark: Output[Dataset]):
    pass


@dsl.component(base_image="python:3.9")
def process_component(data: Input[List[Dataset]]) -> bool:
    import random

    return random.choice([True, False])


@dsl.component(base_image="python:3.9")
def watermark_component(watermarks: Input[List[Dataset]]):
    pass


@dsl.pipeline(name="test")
def pipeline():
    with dsl.ParallelFor(items=[1, 2, 3]) as num:
        sources = source_component(num=num)

    processed = process_component(data=dsl.Collected(sources.outputs["data"]))

    with dsl.If(processed.outputs["Output"] == True):
        watermark_component(watermarks=dsl.Collected(sources.outputs["watermark"]))


if __name__ == "__main__":
    compiler.Compiler().compile(pipeline_func=pipeline, package_path="pipeline.yaml")

when I run

$ python -m test_pipeline

it produces the traceback

Traceback (most recent call last):
  File "/opt/homebrew/Cellar/[email protected]/3.9.16/Frameworks/Python.framework/Versions/3.9/lib/python3.9/runpy.py", line 197, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/opt/homebrew/Cellar/[email protected]/3.9.16/Frameworks/Python.framework/Versions/3.9/lib/python3.9/runpy.py", line 87, in _run_code
    exec(code, run_globals)
  File "/Users/****/work/experiment/test_pipeline.py", line 25, in <module>
    def pipeline():
  File "/Users/****/Library/Python/3.9/lib/python/site-packages/kfp/dsl/pipeline_context.py", line 65, in pipeline
    return component_factory.create_graph_component_from_func(
  File "/Users/****/Library/Python/3.9/lib/python/site-packages/kfp/dsl/component_factory.py", line 673, in create_graph_component_from_func
    return graph_component.GraphComponent(
  File "/Users/****/Library/Python/3.9/lib/python/site-packages/kfp/dsl/graph_component.py", line 68, in __init__
    pipeline_spec, platform_spec = builder.create_pipeline_spec(
  File "/Users/****/Library/Python/3.9/lib/python/site-packages/kfp/compiler/pipeline_spec_builder.py", line 1922, in create_pipeline_spec
    build_spec_by_group(
  File "/Users/****/Library/Python/3.9/lib/python/site-packages/kfp/compiler/pipeline_spec_builder.py", line 1387, in build_spec_by_group
    subgroup_component_spec = build_component_spec_for_group(
  File "/Users/****/Library/Python/3.9/lib/python/site-packages/kfp/compiler/pipeline_spec_builder.py", line 715, in build_component_spec_for_group
    component_spec.input_definitions.parameters[
TypeError: 'NoneType' object cannot be interpreted as an integer

The environment I used

python==3.9.16
kfp==2.8.0

thank you for the help!