This is more of a singer question but how do people usually deal with json/jsonb columns from postgres to

target-s3-csv

? , it seems like it gets unnested with the first key as the json as the column name. An example is there’s a column called

metadata

with each row having different kv. first row being {{

{"foo":"bar"}

next being

{"foofoo":"barbar", "foo":"bar"}

the column in is unnested and turned out to be

metadata__foo

, and value for the two rows are

bar

and

barbar

(For the record, this question was cross-posted to the Singer workspace: https://singer-io.slack.com/archives/C2TGFCZEV/p1595513205349000 🙂)

Oops, thanks. I wrote the post here first and then thought I should cross-post it too just to reach out to more people