In summary, SQL is a better choice than Flux for most use cases in time-series databases due to its full support, extensibility, and compatibility with existing tools and knowledge. While Flux provides greater user control through its flow-based functional model, it also constrains the power of the database and query planner, requiring users to become experts or give up on optimizing query latency. The trade-offs involved in adopting a custom query language like Flux make SQL a more reliable and performant choice for time-series data, particularly when using TimescaleDB, which supports full SQL.