Hey folks, I’m running into something odd when using apollo client with large media-heavy graphql responses. When I pull in project data for things like pc video editing, the cache seems to bloat quickly and sometimes stale fragments get reused, which leads to mismatched UI states after mutations.
Has anyone dealt with cache invalidation or pagination strategies for big binary-style metadata? Also curious if there are best practices to keep memory usage and network traffic under control when the data model gets this heavy.