Hello kedro team, I'm actually working on project of Churn prediction and i finish all pipeline and the job work fine. I work on Kedro 0.16.5 because it was compatible with some packages on our environment. After packging the project now i'm able to run it from command line with:

python3 -m project_name.run

But i have question about parameters. When i run the packaged project i can't anymore pass parameters to the project or modifiy the parameters.yml so my question is how to pass arguments when i run a packaged kedro project ?

so this is much nicer in forthcoming 0.19.x release but you would have to inject them into the packaged wheel somehow on this version

Ok I found out I accidentally muted the channel notification😂 I like this question! @datajoely is correct that in 0.19.x it would be even easier to dynamically change the config location. But you can still work with changing parameters in older versions. I cannot guarantee it works

0.16.5

since this is really old, more than 2 years. It may be surprising but

python -m module

actually accepts the same arguments as

kedro run

, this isn’t documented clearly. TL;DR; So if you can do

kedro run --params=key:value

, you can do the same with

python -m custom_pacakge --params=key:value

to override runtime parameters. The gotcha of the current package project is that. During development, your

conf/base/parameters.yml

is inside the root directory, however, when you move to package mode the parameters have to move into

src/conf/base/parameters.yml

or you have to do some tricks in

settings.py

to read config from somewhere else.

So i tried the

python -m custom_pacakge --params=key:value

but it didn't work for me. And for the file

parameters.yml

i set it in

conf/base/parameters.yml

but in another machine different from the one used before packaging.

Sorry, I don’t have a quick solution in

0.16.x

since the codebase is quite different. I test it with

0.18.3

and I expect this will work for all

0.18.x

series or maybe even older versions.

Finally, i was able to use parameters but by passing them as environment variables and use the TemplatedConfigLoader in the Hook

def register_config_loader(self, conf_paths: Iterable[str]) -> ConfigLoader:
    return TemplatedConfigLoader(conf_paths,
                                 globals_pattern="*globals.yml",
                                 globals_dict={
                                     "date_process": datetime.now().strftime("%Y%m%d%H%M%S"),
                                     "MODEL_OUTPUT": os.getenv("MODEL_OUTPUT"),
                                     "PROJECT_DIR": os.getenv("PROJECT_HDFS_DIR"),

                                 })

and by using the global.yml file i can change some inputs depending if i'm in dev or prod env