Hello, I recently wrote an article [1] about how to use custom external indexes and avoid reparsing Parquet footer metadata in each query.
While this email is definitely a bit of self promotion, I think the blog is relevant to the discussions about adding new indexes and footer structures to the format, as it addresses common misconceptions of what is possible with the current definition. Thank you for your time, Andrew p.s. Our previous post [2][3] was about embedding such indexes into the files themselves [1]: https://datafusion.apache.org/blog/2025/08/15/external-parquet-indexes/ [2]: https://lists.apache.org/thread/81zgxbtxcgk4zyjzs352l5b56kz4b36o [3]: https://datafusion.apache.org/blog/2025/07/14/user-defined-parquet-indexes/
