Snowflake connection

At a glance

The community member is using SQLAlchemy and JupySQL in their Jupyter notebook to run queries, and they are looking for a way to set up SQL cells similar to the '%%sql' notation. Another community member suggests using DuckDB for SQL support and provides a link to a tutorial. The community members discuss using Snowflake to read data, produce charts, reports, and dashboards. The community member also mentions that they have started work on a related feature, but there is still more frontend work and parsing changes needed. In the meantime, the community member suggests using raw SQL with engine.execute() or creating a small wrapper function, and provides some example code.

Useful resources

sshan9124

Hey,

I've read the documentation and searched through existing issues but didn't find what I need.

Currently in my Jupyter notebook I use SQL alchemy and jupysql to run queries in my notebook(see attached)

Is there a way for me to set this up so I can use SQL cells similar to '%%sql' notation
?

7 comments

MMyles Scolnick

can you share a small snippet of what it looks like using SQL alchemy with %%sql?

we support SQL via duckdb (but will extend to other drivers), to learn more about the SQL support, run:

Plain Text

uv pip install "marimo[sql]"
mairmo tutorial sql

sshan9124

Sorry I thought id attached the screenshot

sshan9124

I've used Current timestamp as en example but I can connect to tables

MMyles Scolnick

awesome, this is a great example for me to work towards. we want to support different engines, so we definitely will support this in the med-term.

how are you using snowflake out of curiosity? are you trying to explore snowflake DBs, moving/transforming data around between tables (write-heavy), or trying to show snowflake data in an app (read-heavy)?

sshan9124

The last one, reading data and analysing. Producing charts, reports and dashboards that I could share company wide

MMyles Scolnick

gotcha, thanks for sharing.

i started this work: https://github.com/marimo-team/marimo/pull/3563
but there is a fair amount of frontend work left, and likely some parsing changes to the DAG.

in the meantime, you can use raw sql with engine.execute() or maybe create a small def execute wrapper like something similar similar to the linked PR.

there are also a few examples in there that might be helpful too
marimo/_smoke_tests/sql/polars_sql.py
marimo/_smoke_tests/sql/sql_alchemy_engine.py (only half works on main, other half depends on the PR)e

sshan9124

Missed this, I don't come on discord that often. But this looks interesting I'll have a play

Add a reply

Get help from the marimo community

Snowflake connection