<!channel> Office Hours start in just under an hour! Join us today to chat about SDK updates or anything else that's on your mind. We'll drop a link to the Zoom call when we go live.

Missed this! Sorry 😄 here's my 2 cents fwiw DB Types streams - Couple points, binlog is nice but not every DB has it so don't fret too much! @edgar_ramirez_mondragon - Pydantic / SQLAlchemy stuff. Great hearing you talk about how you're thinking about things!!! @aaronsteers / @edgar_ramirez_mondragon listened deeply to the SQL Alchemy stuff vs direct/odbc. How painful would ripping SQL Alchemy out be if it turns out to be the wrong choice is really where my brain goes. I think I get what you are after. The proprietary type issues are a pain especially depending on which version of JSON Schema you can use. Where are the type hints going to be placed? Having access to the catalog if you're a target can be pretty useful, but at least having some way of mapping types between DBs is useful in some DB -> DB data moves. Looks like the rest of the office were cut off

Where are the type hints going to be placed? Having access to the catalog if you're a target can be pretty useful, but at least having some way of mapping types between DBs is useful in some DB -> DB data moves.

The singer spec supports a

sql-datatype

metadata field (https://github.com/singer-io/getting-started/blob/master/docs/DISCOVERY_MODE.md#metadata). Do you think something like would help in keeping data types more faithful when doing db->db EL?

There’s a second video on YouTube with the next section. They’re still processing but I’m going to try and combine them

https://github.com/singer-io/getting-started/blob/master/docs/DISCOVERY_MODE.md#metadata yes, and the SCHEMA row that's passed to the target doesn't contain that information. I believe what you are showing is just about the catalog. Maybe I"m mistaken? @edgar_ramirez_mondragon