The cloud-native orchestrator for the whole development lifecycle, with integrated lineage and observability.

dagster

If I wanted to pass two config params to all assets in a job, called `start_date` and `end_date`, how would I configure that as the `config` param for the job without specifying it for each asset? Also, how would I specify it in a job so that the end user was required to enter the values in Launchpad before running the job?
```my_job = define_asset_job(
    name="Adhoc_Stuff_Updating_Job",
    selection=AssetSelection.groups(MY_ASSETS),
    config= # ? What goes here?,
)```


hi <@U04M8B4R7NU>! the direct answer to that question is that you could specify that each asset has a required_resource_key of "my_values", then supply the <https://docs.dagster.io/_apidocs/resources#dagster.make_values_resource|make_values_resource> as the implementation of that resource in your Definitions object, i.e.:
```defs = Definitions(..., resources={"my_values": make_values_resource(start_date=str, end_date=str)) ```
then, any job that includes assets with that required resource key will require the user to supply config values for start_date and end_date before launching the job

the less direct answer is that you might want to consider using partitions for this use case. you can declare assets as time-partitioned, e.g.
```@asset(partitions_def=DailyPartitionsDefinition(start_date="2023-01-01"))```

Thanks, Owen. There are some cases where partitions work but I also have need to support more ad-hoc (strange partitioning) jobs. I think your first suggestion would be good! I was wondering whether it could be specified in the job so that I could create multiple jobs that provided different time range via partitions. Monthly, Quarterly, Calendar Year, Fiscal Year, etc. without needing to change the assets for each -- just the job definition.

I'm still sorting through how this all work. :slightly_smiling_face:

would all of those assets be writing to the same location? or would they just be doing similar things with different aggregation windows?

if the second, then i'd recommend an <https://github.com/dagster-io/dagster/discussions/11045|asset factory> pattern, where you can make multiple copies of a similar asset with different partitions_defs (they'll have distinct keys)