How can I prefix the inputs from assets in the function definition by using load_assets_from_modules? Currently it only prefixes the asset itself but not the upstream assets

When using

load_assets_from_modules

to load assets, the function allows you to specify a

key_prefix

that will be prepended to the keys of the loaded assets. However, based on the documentation, there is no direct parameter provided by

load_assets_from_modules

that automatically prefixes the keys of upstream assets that are inputs to the loaded assets. To ensure that the upstream assets have the correct prefixed keys, you would need to handle the prefixing manually. This means that when defining your assets, you should explicitly set the

key

parameter for each

AssetIn

to include the desired prefix. Here's an example of how you might define an asset with prefixed input keys:

python from dagster import asset, AssetIn  @asset(     ins={         "input_asset": AssetIn(key=["my_prefix", "input_asset_name"])     } ) def my_asset(input_asset):     # Your asset logic here     ...

In this example,

input_asset

is an upstream asset, and we are explicitly setting its key with the prefix

"my_prefix"

. If you are loading assets from a module and want to apply a prefix to both the assets and their upstream dependencies, you would need to ensure that the assets are defined with the appropriate prefixed keys as shown above. Then, when you use

load_assets_from_modules

, the

key_prefix

you provide will apply to the assets being loaded, and the upstream assets will already have the correct prefixes as defined in their

AssetIn

specifications. If you encounter issues where the upstream assets' keys are not being prefixed as expected, this might be a limitation or a bug in the current implementation of

load_assets_from_modules

. In such cases, you may want to check the GitHub Issues for Dagster to see if there are any known issues or feature requests related to this behavior.