petere/pgbloomagg

The README doesn't go into detail, but the repository name and description tell the core story: pgBloom adds a "bloom filter aggregate function" to PostgreSQL databases. In practical terms, this is a tool that lets a database quickly figure out whether something is probably present in a large set of data, or definitely not present, without scanning every single row. That can make certain queries much faster, especially when you are working with large tables and just need a fast yes-or-no answer. A bloom filter is a compact data structure that trades certainty for speed and space. Instead of storing the actual data, it stores a small summary that can answer "is this item in the set?" with either "maybe" or "definitely not." When you use it as a database aggregate function, you can build this summary across many rows at once. This is useful for tasks like filtering, deduplication, or pre-checking whether a value exists before doing more expensive lookups. The tradeoff is that a bloom filter can occasionally produce a false positive (saying something is present when it actually isn't), but it will never miss something that is genuinely there. The people who would reach for this are database administrators, backend developers, or data engineers working with PostgreSQL at scale. For example, if you have a large user-events table and want to quickly check whether a particular session ID has ever appeared, a bloom filter can give you a fast answer without scanning millions of rows. It is also useful in analytics pipelines or caching layers where you want to skip unnecessary work. The project is written in PL/pgSQL, which is PostgreSQL's built-in procedural language. Since the README is otherwise empty, the best way to understand how to install and use it would be to look at the code files in the repository directly.