qwen-code bug: server-reported context ceiling lacks scope, can't safely clamp a separate compaction model
yiliang114 · ghdev · 2026-10-11
Issue #13876 in QwenLM/qwen-code documents a compaction flaw: PR #13788 threads the server-reported token limit from context-overflow errors into reactive compaction sizing, but parseTokenCounts returns no scope — no route, model, or whether the bound covers input or the whole window. Two of its six grammars (including DashScope's rangeMatch) yield input-only ceilings, attested by a production [1, 196608] error. The PR therefore clamps only the main-model path; doing it correctly requires carrying scope from the producer. The issue also traces how computeCompactionOutputBudget's 20k-output deduction and 1,024 safety margin floor the budget and trigger COMPRESSIONFAILEDOUTPUTTRUNCATED.
More from coding & agent
- Index Network open-sources Social Discovery Protocol: agents negotiate fit on your behalf — carsonfarmer · 2026-10-23
- Installed a formal-verification skill: it silently installed Java, MacBook runs 10 degrees hotter — zealcaiden · 2026-10-23
- OpenAI's Decisions API hits GA: near real-time routing up to 10x faster than GPT-6 Luna — OpenAIDevs · 2026-10-23
- Ruff author warns: upgrade Node/Bun now — --use-system-ca still accepts the eMudhra cert — charliermarsh · 2026-10-23
- Proof of Agent: paste one prompt and your AI mines, grades, and self-governs — adamamcbride · 2026-10-23
- DeepLearning.AI weekly: GPT-6.1 hits 97.9% byte parity on COBOL migration — DeepLearningAI · 2026-10-23