What problem does it solve? When you need concurrent task execution in Python without deploying Redis, RabbitMQ, or other middleware, traditional ThreadPoolExecutor lacks QPS rate limiting, retries, timeouts, and result retrieval. This Skill guides you to use funboost's MEMORY_QUEUE broker and pool classes as a zero-dependency local concurrency solution. ## Core Features & Use Cases - ThreadPoolExecutor Replacement: Swap ThreadPoolExecutor(...) for MemoryFunboostPool(...) or FunboostPool(...) with an identical pool.submit(fn, *args) API, gaining QPS control and concurrency management. - In-Process Future Results: Use get_future() and get_aio_future() to retrieve task results via concurrent.futures.Future without any Redis RPC dependency. - Non-Serializable Arguments: Pass arbitrary Python objects (database connections, custom class instances) as task arguments since memory queues skip serialization entirely. - Use Case: During local development, run tasks with BrokerEnum.MEMORY_QUEUE; when deploying to production, change only the broker_kind line to REDIS_ACK_ABLE while keeping all business code unchanged. ## Quick Start Ask the AI to rewrite your ThreadPoolExecutor-based concurrent code using funboost MemoryFunboostPool with a specified concurrency number and QPS limit.
