hckrnws

Postgres’s extensible index AM story doesn’t get enough love, so it’s nice to see someone really lean into it for LIKE. Biscuit is basically saying: “what if we precompute an aggressive amount of bitmap structure (forward/backward char positions, case-insensitive variants, length buckets) so most wildcard patterns become a handful of bitmap ops instead of a heap scan or bitmap heap recheck?” That’s a very different design point from pg_trgm, which optimizes more for fuzzy-ish matching and general text search than for “I run a ton of LIKE '%foo%bar%' on the same columns”.

The interesting question in prod is always the other side of that trade: write amplification and index bloat. The docs are pretty up-front that write performance and concurrency haven’t been deeply characterized yet, and they even have a section on when you should stick with pg_trgm or plain B-trees instead. If they can show that Biscuit stays sane under a steady stream of updates on moderately long text fields, it’ll be a really compelling option for the common “poor man’s search” use case where you don’t want to drag in an external search engine but ILIKE '%foo%' is killing your box.

Looks very interesting. I really like trigram indexes for certain use cases, but those are essentially running an ILIKE %something% on various text content in the DB. So that would fit the described limitations of this index type very well.

Usually you're quickly steered towards fulltext search (tsvector) in Postgres if you want to do something like that. But depending on what kind of search you actually need, trigram indexes can be a better option. If you don't search so much for natural language, but more for specific keywords the stemming in fulltext search can get in the way.

One information that would be nice here is a comparison of the index size on disk for both index types.

How is the postgres ecosystem at stating when these kinds of things are ready for adoption? I can think of a usecase at work where this might be useful, but hesitant to just start throwing random opensource extensions at our monolith DB.

This is a fairly simple idea of indexing characters for each column/offset and compressing the bitmaps. Simple is good, as the overhead of more sophisticated ideas (eg suffix sorting) is often prohibitive.

One suggestion is to index the end-of-string as a character as well; then you don't need negative offsets. But that turns the suffix search into a wildcard type of thing where you have to try all offsets, which is what the '%pat%' searches do already, so maybe it's OK.

Would this be a good fit to replace FTS for hybrid search? Biscuit + Vector?

Any data on index size for big tables? Comparison (with ms/megabytes) vs trigram regarding size/speed?

UPD

> Biscuit is 15.0× faster than B-tree (median) and 5.6× faster than Trigram (median)

> Trade-off: 3.2× larger index than Trigram, but 5.6× faster queries (median)

I found some more info here: https://biscuit.readthedocs.io/en/latest/benchmark_roaring.h...

Noticed Daniel Lemire talking about it and how they use Roaring Bitmaps.

https://x.com/lemire/status/2000944944832504025