datahike 2026-08-04

I fixed a bunch of reported issues such as missing math functions in https://github.com/replikativ/pg-datahike, and improved Datahike's support for cached prepared statements/queries, which significantly improved pg-datahike's performance vs. Postgres on pg-bench. pg-datahike is now only up to ~5x slower on throughput, while being horizontally scalable, e.g. in the https://github.com/replikativ/datahike-saas-starter setting.

I am still exploring the performance trade-offs and there might still be classes of queries that are significantly slower. I am happy to get feedback and have people report on what they would need.