`pipeline.pipeline` confuses IDE autocompletion

[noklam]

Description

We import pipeline in our starter and this confuse IDE like VSCode, as it thinks pipeline is a module instead of the pipeline function.

from kedro.pipeline import Pipeline, node, pipeline

Changing it to the following avoid the issue.

from kedro.pipeline.modular_pipeline import pipeline

The problem here is IDE confuse kedro.pipeline.modular_pipeline as kedro.pipeline.pipeline module.

Context

Solution 1:

• Move the pipeline definition (Or what's the difference between pipeline and modular pipeline? Will be a breaking change if we move modules)

Solution 2:

• Change starter to import from modular_pipeline explicitly, should we keep the shortcut in this case?

Solution 3:

• find out ways that won't confuse IDE (VS Code in my case, user report this with PyCharm too)

Expected Result

Actual Result

-- If you received an error, place it here.

-- Separate them if you have more than one.

Your Environment

• Kedro version used (pip show kedro or kedro -V):
• Python version used (python -V):
• Operating system and version:

[astrojuanlu]

or

from kedro import Pipeline

Pipeline.make(...)

😄 #712

(see also #2723 about "modular pipelines" vs just "pipelines")

[noklam]

Added 0.19 tag to prompt for discussion in case we need to move stuff around. (It's annoying Github project change doesn't allow comment🥲)

[noklam]

Discussed in backlog grooming. Will keep this as a low priority for the 0.19 release and see if we have time to fit this in.

If we end up don't want to break anything in 0.19, we should still update the starters to avoid these syntax highlight problem by importing from kedro.pipeline.modular_pipeline directly.

[astrojuanlu]

There's more confusion: https://github.com/kedro-org/kedro-viz/issues/1522#issue-1885805761

Some users are creating the pipelines with the Pipeline class.

I think we should tackle this and #2723 at once.

[astrojuanlu]

I just saw this code snippet in someone else's code:

my_pipeline = Pipeline([
    Node(filter_func, "df_input", "df_filtered"),
    Node(actual_func, "df_filtered", "df_output"),
])

is there a reason why we are using the pipeline and node helpers instead of the Pipeline and Node class initializers directly?