Meilisearch vs SurveyJS

Meilisearch ships a tight two-week release train from a single GitHub tag stream, mixing user-facing engine work with prototype tags. The last three minor releases all revolve around Dynamic Search Rules — an experimental system for changing ranking and filtering behavior per query, time window, or facet value — alongside steady work on sharded, federated deployments. Performance is a parallel constant: synonym storage, result formatting, and search-request paths have all been reworked for lower latency in the last month.

The product is moving from a fast embedded search index toward a programmable relevance layer that operators configure rather than recompile. DSR went from an array of ad-hoc conditions to a typed condition object with query, time, and filter branches in two releases, which reads like an API being hardened toward stability. In parallel, sharded federated fetch and the new SSE task and batch streams point at larger multi-node installs where clients need to observe engine state continuously instead of polling.

Expect DSR to keep accreting condition types and then graduate out of the experimental flag, with the SSE streams following the same path once the task payload shape settles.