Repository navigation
fix(agentx): upgrade harness to 1.0.10 for SPAWN worker affinity - #4119
Conversation
|
Thanks for the contribution!
中文感谢你的贡献！
|
|
Generally most InferenceX submissions don't need to add changes to the infx package. On the off chance they do, generally the diff is small. Please prompt your agent to clean your PR to keep your diff to only necessary & minimal changes too. We have an increased code quality bar for changes to the infx package. 中文大多数 InferenceX 提交通常不需要修改 infx 包。即使确实需要，改动通常也很小。请让你的 agent 清理 PR，使改动只包含必要且最小的内容。infx 包的改动有更高的代码质量要求。 |
![claude[bot]](https://avatars.githubusercontent.com/in/1236702?s=60&v=4){kind=small}
There was a problem hiding this comment.
Looks good, straightforward dependency version bump with doc sync and a properly formatted changelog entry.
What was reviewed:
- Version consistency: requirements.txt and the standalone doc both now pin agentx-harness 1.0.10, no stray 1.0.9 references left behind.
- perf-changelog.yaml entry appended at the physical tail, English-only, correct schema fields.
- Wildcard
config-keys: ["*"]+scenario-type: agentic-codingscoping mirrors existing precedent for harness-wide pin bumps (e.g. the prior srt-slurm pin entry), so it's appropriately scoped rather than overly broad. pr-linkstill shows thepull/XXXplaceholder, but that matches CONTRIBUTING.md's own documented example and two pre-existing changelog entries use the same placeholder, so it isn't a new issue introduced by this PR.
Extended reasoning...
Three-file diff: a version bump from agentx-harness 1.0.9 to 1.0.10 in a doc and a requirements.txt, plus one appended perf-changelog.yaml entry; no code paths, auth, or data handling are touched. Verified the version bump is consistent across both files, the changelog append preserves prior bytes and lands at the physical tail, and the wildcard config-keys usage with scenario-type scoping has direct precedent elsewhere in the file for similar harness-level dependency bumps. The only minor wrinkle (pr-link placeholder pull/XXX) is a pre-existing, documented pattern in this repo rather than something unique to this change, so it doesn't warrant blocking.
This review covers commit e8928bc, which is no longer the latest commit on this pull request; later commits are not covered by it.
…om-disagg-agentx Resolve the perf-changelog conflict with #4065 and #4119 using prepare_perf_changelog_merge resolve-conflict; this PR's entry stays the tail append. 用 prepare_perf_changelog_merge resolve-conflict 解决与 #4065、#4119 的 perf-changelog 冲突，本 PR 的条目仍是末尾追加。 [skip-sweep] Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Upgrade the isolated AgentX benchmark dependency from
agentx-harness==1.0.9to1.0.10and update the standalone install guide. The release includes Anthony's SPAWN worker-affinity fix, preventing child sessions from accumulating duplicate state across client workers. Metric collection, calculations, and exports are unchanged.Release: https://github.com/SemiAnalysisAI/agentx-harness/releases/tag/agentx-v1.0.10. Fix: SemiAnalysisAI/agentx-harness#52.
The fix-only c6656 AgentX run completed with zero profiling errors. Independent reconstruction from all 781,455 profiling records reproduced the core exported statistics and full inter-chunk distribution: https://github.com/SemiAnalysisAI/InferenceX/actions/runs/38683542278/job/116253300866.
The 1.0.10 release passed benchmark smoke tests from wheel and source installations on Python 3.11–3.13, then installed the published package from PyPI and passed the benchmark smoke test again: https://github.com/SemiAnalysisAI/agentx-harness/actions/runs/38764721499. The published worker-router code is byte-identical to the patch used in the c6656 run.