goutamadwant commented on PR #16369: URL: https://github.com/apache/dubbo/pull/16369#issuecomment-5421314581
@zrlw I implemented the requested warm-up-only direction and the focused regression passes 10 fresh JVMs with 200 concurrent decodes per JVM. Before treating it as ready, I found one loader-scope gap that the test does not cover: the warm-up state is keyed only by the requested class, so generic `Object.class` decodes can share a warmed flag across different context classloaders and take the lock-free path while a new reader graph is still cold. The steady path also calls `setCreator` for every decode. Would you prefer the warm-up state to be scoped by context classloader plus target class, with creator selection kept inside the cold path, or is there a fastjson2 readiness signal we should use instead? I have not added another speculative patch pending that direction. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
