What would be the preferred way to connect a jdbc-...

# datascience

[Paul Woitaschek]

What would be the preferred way to connect a jdbc-trino connection with notebooks and utilize dataframes?

[altavir]

I can't find jdbc connections in DataFrame. Our own project tables-kt (https://github.com/SciProgCentre/tables-kt) has connections both to DataFrame and to jdbc (via Exposed). So it could in theory be used as a bridge. Sadly, it is not out of prototype stage and not documented. You can try to use it or take the code from appropriate modules. I can ask questions if you do.

[altavir]

Tables-kt is intended as a connector library, that supports both row-based (SQL) and column-based (DataFrame) representation. But we do not have manpower to polish it. Contributions are welcome.

[Paul Woitaschek]

Thanks! I tried the exposed way and it doesn't support Trino

[altavir]

It is possible to implement a Table for raw JDBC, but it will require some work...

[Ilya Muradyan]

@zaleslaw had some experiments with JDBC recently

[Jolan Rensen [JB]]

Yes, the DF team is currently exploring database connections using JDBC. It's still in its infancy/experimental stage.
You can keep up with this at https://github.com/Kotlin/dataframe/issues/212 or https://github.com/Kotlin/dataframe/issues/408 and feel free to share your usecase there too 🙂

[zaleslaw]

Thanks, we have plans to release jdbc connector, for now, you should read tables manually

[Paul Woitaschek]

Thanks! Exposed doesn’t work for us as it doesn’t support trino.
I shared some hacky code here:

https://github.com/Kotlin/dataframe/issues/212#issuecomment-1686342865