just a disclosure: I haven't gone through the flattening logic here but seems pretty odd, if I remove flattening -- almost a lot of schema errors are gone.

I also feel jira is particularly weird because you can add a custom field and then fetch it, so the contract is somewhat not public which means to: 1. Not have customfields and expect users to add it 2. or have some fields and let users create a fork and use it, however they like. but my problem was other than custom fields and more so than the fields why the flattening messed up the datatypes

backtracking here a bit, I read the code now for flattening and it turns out all is fine, it's just sdk flattening doesn't handle ArrayType and just dumps array as a string (link)

still thinking why flattening fails the schema validation, as from the code I can see flattening applies to both, schema and the record

I am not able to understand why the keys of schema in

IssueStream

in tap-jira are flattened but the values are not updated to string but when I use the flattening helpers it does evaluate to correct version.