Schema inference necessary!

[alanpaulkwan]

I'm very interested in embedding DuckDB into a server context. This is almost that, except I would have to define all Parquet files' columns as I read them? This makes it quite unpalatable to use. Is this an inherent limitation of integrating into Postgres or is there a way around it?

[JelteF]

The next release will make the syntax much easier, and should resolve your issues I think due to #531 for details. The current docker images on dockerhub are outdated though, which I'm trying to fix in #570. If you want to try before I fixed docker, you can compile form source.

[alanpaulkwan]

Awesome thanks! When the docker is fixed I'll give it a whirl.

[alanpaulkwan]

So I can try this now?

[JelteF]

Yes, you can try this now. I verified that the latest docker images contain the new feature.