Hi all, Quick question: I have a run env, here before I run a pipeline I need query some endpoint that returns a parameter dict. Like:

extra_args: dict = client.get_params(run_desc)

Then I would do something like:

actual_params = OmegaConf.merge(catalog.load("parameters"), extra_args)

Where can I best do this? I noticed for example that if I try to modify parameters in a

before_pipeline_run

hook, the catalog params has already been split out as

parameters

,

params:sec1

etc, and if this

extra_args

changes parameters broadly then it's not trivial walk through all and manipulate via the catalog. What's the best way to do this?

after_context_created

is the right hook to use. There are two approaches depending on your use case: If

extra_args

only overrides keys already defined in

parameters.yml

, you can use the

runtime_params

resolver [link] If

extra_args

adds new keys not already in

parameters.yml

, you can bypass the config loader directly. See the docs on bypassing configuration loading rules [link].

Thanks @Rashida Kanchwala Here, I want to both override and add new, so per my understanding,

runtime_params

is the right approach. However, one small question: in

after_context_created

, the

context

has already been instantiated.

runtime_params

is an init arg for

KedroContext

but actually stored as a

_runtime_params

, i.e. implied to be "private". -> Is there any risk or downside to modifying the

_runtime_params

after instantiation?

You can access runtime_params this way instead

context.config_loader.runtime_params

👍