Auxten Wang

Thanks for the excellent suggestions, @cmaureir. I'll work on optimizing disk space usage.

Data seems contain dict type, debuging ```csv id community level ... findings rank rank_explanation 0 e7822326-4da8-4954-afa9-be7f4f5791a5 42 2 ... [{'explanation': 'Marley's Ghost plays a cruci... 8.0 The impact severity rating...

Currently, ClickHouse will read the column "findings" as `Array(Tuple(explanation Nullable(String), summary Nullable(String)))` from parquet. If the [Semistructured Columns](https://github.com/ClickHouse/ClickHouse/issues/54864) is done. Maybe chDB can use JSON type for this kind of...

chDB is ClickHouse, so all things would be like [executable UDF functions](https://clickhouse.com/docs/en/sql-reference/functions/udf) Here is an example: ``` root@0a8b55995b6e:/auxten/chdb/tests# python3 udf3.py "Value 2" ``` udf3.py ```python import chdb ret = chdb.query(...

@djouallah @blackrez Thanks for your long term support and suggestions. We will continue to improve the core functionality of chDB. You may have noticed that in the past six months...

> Any thoughts on https://github.com/vortex-data/vortex? Glad to hear more about your thoughts! 😄 This is the first time I've heard of vortex.

In current version. Connection based API (No.2) is the fastest

PR is here https://github.com/duckdblabs/db-benchmark/pull/131

@blackrez Let me understand your point. Your suggestion is to use the Native output format as the intermediate format between the ClickHouse engine and language bindings of chDB to speed...

There is someone coming to the Pydata booth suggesting clickhouse-connect support chDB as an optional backend.