RAGFlow is an agent-native database: a Retrieval-Augmented Generation (RAG) engine that grew into context infrastructure.
Go agent-harnessagent-native-databaseagentic-aiagentic-retrievalagentic-searchaiai-agentscontext-enginecontext-engineeringcontext-managementharness-engineeringknowledge-compilationragretrieval-augmented-generationretrieval-harnesssearch-harness
fix(agentic-rag): deliver the final answer after a budget stop (#21301)
Caobaining1e2401e54947d
diff --git a/internal/service/agentic_pipeline.go b/internal/service/agentic_pipeline.goindex d3509be86c..c82a676a8c 100644--- a/internal/service/agentic_pipeline.go+++ b/internal/service/agentic_pipeline.go@@ -146,11 +146,23 @@ func (s *ChatPipelineService) agenticRag(runState.Sys["tenant_id"] = chat.TenantID}runCtx = runtime.WithState(runCtx, runState)+ // A result is abandoned only when the CALLER is gone. runCtx is not the+ // delivery signal: the token budget cancels it precisely so that the+ // terminating fallback can answer from the evidence already gathered, so+ // treating that cancellation as "nobody is listening" dropped the very+ // answer the run had just built and shipped an empty reply. Delivery is+ // tried first and unconditionally, because out is buffered and the+ // consumer always drains it.emitResult := func(result AsyncChatResult) bool {select {case out <- result:return true- case <-runCtx.Done():+ default:+ }+ select {+ case out <- result:+ return true+ case <-ctx.Done():return false}}