mirror of
https://github.com/QwenLM/qwen-code.git
synced 2026-08-26 09:04:37 +00:00
* docs(channels): define immediate delivery v1
* feat(channels): add proactive delivery boundary
* feat(daemon): route channel delivery to workers
* feat(scheduler): persist channel delivery targets
* feat(daemon): accept reverse channel deliveries
* feat(daemon): deliver prompt and scheduled finals
* feat(daemon): add synchronous channel notify
* fix(channels): classify Feishu delivery network errors
* fix(daemon): harden channel delivery failures
* fix(channels): wrap DingTalk proactive token fetch in typed delivery error
A proactive token fetch failure escaped sendProactiveChunk as a plain Error, bypassing ChannelProactiveDeliveryError classification. Wrap it as a transient typed error (preserving the cause) so downstream classification dispatches on it consistently. Also cover the scheduled-source branch of the bridge channel-delivery ext-method, which previously had no test coverage.
* refactor(cli): rename shutdown drain constant to reflect dual scope (#7388)
* fix(cli): support delivery updates in scheduled-task PATCH (#7388)
* fix(cli): address review feedback on channel delivery (#7388)
- Guard empty text before scheduling channel delivery in both prompt
and cron paths, avoiding unnecessary IPC round-trips when the model
emits only tool calls
- Import MAX_CHANNEL_DELIVERY_NAME_LENGTH and
MAX_CHANNEL_DELIVERY_TARGET_ID_LENGTH from core instead of
redeclaring them, preventing potential drift between the persistence
and API validation layers
- Remove dead thought parameter from #collectChannelDeliveryText —
all call sites pass a single argument
* fix(cli): add cross-reference and invariant comments for channel delivery (#7388)
Co-authored-by: Qwen-Coder <qwen-coder@alibabacloud.com>
* fix(cli): revert doc corruption and harden delivery event publish (#7388)
* fix(docs): revert prettier-mangled GitHub Action doc and add prettierignore guard (#7388)
* fix(cli): address review feedback on channel delivery (#7388)
- Make deliverChannelMessage error workspace-aware when routing fails
- Discriminate DaemonChannelDeliveryResultData union on status field
- Add old-target rejection assertion to PATCH delivery test
- Add IPC send failure path tests for channel delivery
- Add rehydration → authorization store integration test
* docs(cli): clarify delivery text scope, consume tradeoff, and 503 expectation (#7388)
* fix(cli): address review feedback on channel delivery (#7388)
- Document single-turn invariant on channelDeliveryCollector field
- Note adapter disposition drop in design doc for V1 scope
- Add text length bound to IPC request validator (defense-in-depth)
* docs(daemon): define final-only channel delivery
* fix(daemon): deliver only final response blocks
* test(daemon): cover cancelled scheduled delivery
* fix(daemon): skip delivery for unbound tasks, consolidate error codes (#7388)
* fix(feishu): validate message response codes
* Revert "fix(feishu): validate message response codes"
This reverts commit 85fc0943d69d2f5e7535477f2d46bfcd8c5c3c1e.
* refactor(channels): defer DingTalk error classification
* fix: apply CR suggestions from AI review
* test(channels): cover delivery failure paths
* fix(cli): add required ink FrameCell/ReadonlyFrame properties to selection tests
The branch was behind main where
|
||
|---|---|---|
| .. | ||
| plans | ||
| specs | ||