How can I log markdown-based metadata when working with sofware-defined-assets? I.e. In the manual asset way of doing things:

yield MetadataEntry.md(docstring_schema, "DataFrame columns")

https://docs.dagster.io/concepts/ops-jobs-graphs/op-events can nicely be used to manually materialize the data and then manually log the metadata entry (which supports the markdown type). However when SDA are used the return value is the dataset and

context.add_output_metadata

can add (string) based metadata. JSON seems to be recognized automatically - but MD not. Trying to directly feed it:

context.add_output_metadata({'df columns': MetadataEntry.md(docstring_schema, "Dataframe Columns")})

fails with: DagsterInvalidMetadata: Expected a metadata value, found an instance of MetadataEntry. Consider instead using a MetadataValue wrapper for the value

This should be

context.add_output_metadata({"Dataframe Columns": MetadataValue.md(docstring_schema)})

(basically, using

MetadataValue

instead of

MetadataEntry

). The reason that MetadataEntry does not work is that this class contains both the name of the field as well as the value, but when you pass in a dictionary of metadata, the keys already represent the field names.

thanks