We've had trouble with

meltano elt

runs failing when precision doesn't match among the tap, target, and Python. For example, we had tap-mongodb producing a precision of 1e-34, while the default precision of Python's

decimal

library is 28. Targets that verify the schema fail when this happens. We looked at how one of the target-postgres solved this, and encapsulated it into a standalone repository https://github.com/Datateer/adjust-precision-for-schema. I hope this is useful for target developers Using it is simple. This function walks the schema, inspecting it for numeric precision, and adjust's the Python context's precision if needed. An example is at https://github.com/ome9ax/target-s3-jsonl/pull/20/files#diff-e73ec1af3a3ffa6444c10840dea2dbefa7aed8ba8f256548f334a510a4f2e856R230

This is great! Thanks for sharing. I don't recall if the SDK does something like this already but regardless, it is very nice to have this as a standalone importable library. 🚀

I learned the newer versions of the SDK does transform numeric fields into

decimal.Decimal

instances. Which is nice. But it doesn't handle

decimal.DivisionImpossible

exceptions. This does