Kimmo Linna

The iPadOS version is 15.7 which is the latest my employer has approved so I assuma my Safari version on iPad is then 15.

I have understood that at this point you could fetch data to an ArrowTable but if you look [DuckDB-header](https://github.com/duckdb/duckdb/blob/master/src/include/duckdb.h) file there isn't possibility to do like this Python code does...

But I would say that @hannesmuehleisen is a right person to point us to right direction with this...

Why would you use Arrow Table for queries when you can use DuckDB? Quote from arrow.apache.org: "DuckDB’s state-of-the-art optimizer can push down filters and projections directly into Arrow scans. As...

@asbisen What would be the most common way to this? DBInterfaces.jl or Tables.jl?

I have noticed that too. DBInterface should also support columnwise insert for example to DataFrame to get the best out from the column oriented database.

@mytherin Could you tell us what is the status at the moment? 😉

@tim-salabim I think that this is the limitation of duckdb-wasm at the moment. When I try `*.parquet` in `shell.duckdb.org` I will get the following error message: `Invalid Error: [object WebAssembly.Exception]`...

@ankoh Because the Content Security Policy doesn't support the wildcard * in the path, [match-paths spec](https://www.w3.org/TR/CSP3/#match-paths) could it possible that instead of `s3://bucket/folder/*.parquet` I could use `s3:/ /bucket/folder/` that means...

@tim-salabim One solution is use AWS-SDK to fetch file names and then use `read_parquet(list of files)` [My test ObservableHQ Notebook](https://observablehq.com/@kimmolinna/duckdb-parquet-and-aws-s3 )