Query, Mutation, and Stream Are Not the Same Kind of Async Work

Short summary

The article argues that frontend async work should not be flattened into a single fetch-setState pipeline. Queries, mutations, and streams have fundamentally different requirements around identity, caching, invalidation, lifecycle, and consistency. Treating them as distinct semantic concepts with proper ownership in a reactive graph prevents the state confusion that plagues most frontend architectures.

• •Queries need identity and should be resource nodes, not component-local state
• •Mutations have global impact but logic is often scattered locally across components
• •Streams differ from both in lifecycle and data ownership semantics

Generated with AI, which can make mistakes.