I am using the documented sensor paradigm which monitors an s3 bucket for new files and invokes a job that materializes an asset. The problem I have is that the output produced is the same for each different source file and I can't figure how to make each invocation of the asset yield a different asset key per source file given to it. Ideas?

when you define an asset in software, the same asset key is intentionally given to every artifact materialized by that software defined asset. This is probably better modeled as a partition of the asset per file: https://docs.dagster.io/concepts/partitions-schedules-sensors/partitions#dynamically-partitioned-assets

@chris So do the emitted outputs have different path keys when partitioning is used so that the iomanager can save them to different places?

that’s right

thanks @chris