I just downloaded duckdb and used it to query parquet files. It's amazing. Finally, an easy way to query Parquet data without spinning up a local Spark instance or doing pd.read_parquet. It's pretty fast too.
I've always felt that Parquet (or ORC) ought to be more popular mainstream formats than they are, but in practice the interactive tooling for Parquet has always been weak. For years, there wasn't a good GUI viewer (well, there were a few half-hearted attempts) or a good SQL query tool for Parquet like there is for most relational databases, and lazily examining the contents of a large Parquet folder from the command line has always been a chore. At least now, DuckDB's parquet support moves the needle on the last point.
I've always felt that Parquet (or ORC) ought to be more popular mainstream formats than they are, but in practice the interactive tooling for Parquet has always been weak. For years, there wasn't a good GUI viewer (well, there were a few half-hearted attempts) or a good SQL query tool for Parquet like there is for most relational databases, and lazily examining the contents of a large Parquet folder from the command line has always been a chore. At least now, DuckDB's parquet support moves the needle on the last point.