tjgreen

A little birdie told me that efforts are underway to support the extension in Alloy, at least!

There is indeed such a tradeoff. The architecture is designed with an eye towards making this tradeoff tunable (frequency of memtable spills, aggressiveness of compaction) but the work here is not yet finished. We chose to prioritize optimizing bulk-indexing and query performance for GA, since this is already enough for many applications. I'm excited to get to the point where we have brag-worthy benchmark numbers for high-frequency updates as well!

Yes, hybrid search is one of the main current use cases we had in mind developing the extension, but it works for old-fashioned standalone keyword-only search as well. There is a lot of art to how you combine keyword and semantic search (there are entire companies like Cohere devoted to just this step!). We're leaving this part, at least for now, up to application developers.

You'll have to ask Supabase!

For 2M scale dataset, you should be able to index in about 1 minute on low-end hardware. See the MS-MARCO v1 (8M documents) numbers, measured on cheap Github runners.

Yep, there are numbers in the blog post and repo. We are able to index MS-MARCO v2 (138M documents, around 50GB of raw data) in a bit under 18 minutes.

Okay then!

I actually don't love this example either, for the reasons you mention, but at some point we had questions about how to filter based on numeric ranking. Thanks for the reminder to revisit this.

Re filtering, there are often reasonable workarounds in the SQL context that caused me to deprioritize this for GA. With your example, the workaround is to apply post-filtering to select just matches with all desired terms. This is not ideal ergonomics since you may have to play with the LIMIT that you'll need to get enough results, but it's already a familiar pattern if you're using vector indexes. For very selective conditions, pre-filtering by those conditions and then ranking afterwards is also an option for the planner, provided you've created indexes on the columns in question.

All this is just an argument about priorities for GA. Now that v1.0 is out, we'll get signal about which features to prioritize next.