Another question wrt `kedro-boot`: how is do the

AppPipeline

objects work with

find_pipelines

? my

AppPipeline

is found by find_pipelines, but then when in the apps route i do:

kedro_boot_session.run(
            name="web_api",
…)

the view cannot be found. only when i do

kedro boot --app … --pipeline web_api

, my

AppPipeline

is included in the kedro_boot_session._pipeline.views

Your

find_pipeline

return a dict with multiple pipelines including the

web_api

pipeline. When you run your app with

__default__

it will not include your

web_api

pipeline, and subsequently not including your

web_api

pipeline view.

Kedro boot

use the native

kedro run

behavior regarding pipeline selection and filtering through CLI. The only pipeline that is exposed to your fastapi app, is the currently selected pipeline from your

kedro boot

command (

__default__

pipeline by default). If you want to expose multiple pipeline views to your apps, you can append multiple

app_pipeline

and register them in a single entry (

__default__

for example)

But the default

register_pipelines

function includes:

pipelines["__default__"] = sum(pipelines.values())

So shouldnt the

AppPipeline

be included in the view? Or does kedro-boot only allow one to expose the pipeline if the entire pipeline is of type

AppPipeline

? I dove quite deep into the code through debugging, but did not get to the part where the pipelines where added the views. Anyway, would that not be desired behaviour? Or should it not at least be documented, that if you dont add your

AppPipeline

as the default, you have to pass the --pipeline flag?

Pipeline + AppPipeline + …= Pipeline AppPipeline + Pipeline + … = AppPipeline The

sum(pipelines.values())

likely sumed a

Pipeline

with an

AppPipeline

which return a Regular

Pipeline

as default this is why you dont find your

web_api

view. A regular

Pipeline

is also exposed to the app and can be selected with

kedro_booot_session.run(name=´__default__’)

or by not giving a name

kedro_booot_session.run()

. This can be useful for quickly triggering the selected

Pipeline

without declaring an

AppPipeline

, but it will not expose the inputs, parameters and artifacts. Only

template params

that can be used with a regular

Pipeline

It is true that we need to document this behaviors in a proper documentation site. Your feedbacks will help us to guide the upcoming documentation and features.