I have another question about using partitions with assets. What's the partition field that can be accessed inside the asset?

context.partition_key

?

to clarify, are you asking what attribute to get off the context to find the partition? or what value

context.partition_key

will have?

yeah just what attribute

yeah

context.partition_key

will work. you can also use

context.asset_partition_key_for_output(<output_name>)

oh, is that so that a different asset can still access the partition key for an upstream asset?

i’m not totally sure - @claire can you confirm?

Ah, the function that will give you partition key of the input that maps to the current asset is the

asset_partition_key_for_input

function. Both methods that jamie mentioned pretty much do the same thing for SDAs--fetching the partition key for the current asset. The

asset_partition_key_for_output

method exists separately because some users use a pre-SDA experimental feature where a partitions definition exists directly on an output.

oh ok great, thanks!

ah, glad it worked! sorry about the late response--the documentation for the

asset_partition_key_for_input

is here: https://docs.dagster.io/_apidocs/execution#dagster.OpExecutionContext

context.partition_key

will return the output partition key of your asset.

asset_partition_key_for_input

will return partition key of the upstream input that the current asset depends on. So for example, if your current asset is an hourly-partitioned asset that depends on a daily-partitioned asset,

context.partition_key

will return an hourly-partitioned key, while

asset_partition_key_for_input

will return a daily-partitioned key.

thank you!