Is there a good pattern for configuring destination column types with the

target-redshift

repos that meltano endorses? Or at least graceful upgrading of columns over time? Context: previously we were using https://github.com/transferwise/pipelinewise-target-redshift - which makes an interesting assumption that all numeric columns should get persisted as

double precision

. This isn't exactly ideal for financial information - which is the use case we're solving for. I took a look at the version that's maintained https://github.com/TicketSwap/target-redshift - and it looks like it also makes the same assumption. https://github.com/TicketSwap/target-redshift/blob/main/target_redshift/connector.py#L237

Hi @Michael Sandt 👋

Is there a good pattern for configuring destination column types with the

target-redshift

repos that meltano endorses? Or at least graceful upgrading of columns over time?

Yes, there is. That's what the schema _extra_ is for. The target is unfortunately not on the singer-sdk 0.45.x version, but if you are able to fork their repo to bump it, then you could use the

x-singer.decimal

string format:

extractors:
- name: <tap-yourtap>
  schema:
    <some_stream_id>:
      <some_numeric_field>:
        type: ["string", "null"]
        format: x-singer.decimal
        scale: ...
        precision: ...