fix: stabilize uv lock after experimental changes

[terrykong]

What does this PR do ?

Address instability introduced by this PR #1334

Errors users saw before this:

× Failed to download and build `deep-ep @
  │ git+https://github.com/deepseek-ai/DeepEP.git`
  ╰─▶ Extra build requirement `torch` was declared with `match-runtime
      = true`, but `deep-ep` does not declare static metadata, making
      runtime-matching impossible

when changing unrelated delepdencies

Issues

List issues that this PR closes (syntax):

Usage

• You can potentially add a usage example below

# Add a code snippet demonstrating how to use this

Before your PR is "Ready for review"

Pre checks:

• Make sure you read and followed Contributor guidelines
• Did you write any new necessary tests?
• Did you run the unit tests and functional tests locally? Visit our Testing Guide for how to run tests
• Did you add or update any necessary documentation? Visit our Document Development Guide for how to write, build and test the docs.

Additional Information

• ...

Summary by CodeRabbit

• Chores
  • Updated dependency version specifications for causal-conv1d, mamba-ssm, deep_ep, and deep_gemm to ensure consistent version pinning across configuration.

[coderabbitai]

📝 Walkthrough

Walkthrough

Configuration file pyproject.toml was updated to reformat the no-build-isolation-package entry and add explicit version pinning for four dependencies (causal-conv1d, mamba-ssm, deep_ep, deep_gemm) across two duplicated dependency metadata sections.

Changes

Cohort / File(s)	Change Summary
Configuration & Dependency Metadata pyproject.toml	Reformatted no-build-isolation-package list from single-line to multi-line format; added explicit version fields for causal-conv1d (1.5.0.post8), mamba-ssm (2.2.4), deep_ep (v1.1.0+e3908bf), and deep_gemm (v2.0.0+7b6b556) in both tool.uv.dependency-metadata sections for consistency.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

Pre-merge checks and finishing touches

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	You can run @coderabbitai generate docstrings to improve docstring coverage.

✅ Passed checks (3 passed)

Check name	Status	Explanation
Title Check	✅ Passed	The title "fix: stabilize uv lock after experimental changes" directly relates to the changeset, which adds explicit version fields to dependency metadata in pyproject.toml and ensures consistency across duplicated sections. These explicit version pins serve to stabilize the uv lock file, making the title an accurate reflection of the PR's primary purpose. The title is specific, concise, and uses the "fix:" prefix appropriately for a stability improvement. A reviewer scanning the history would clearly understand that this PR addresses lock file stabilization by pinning dependency versions.
Test Results For Major Changes	✅ Passed	The PR changes consist of reformatting and adding explicit version fields to dependency metadata in pyproject.toml. These are minor maintenance changes focused on stabilizing the uv lock file by pinning versions of dependencies (causal-conv1d, mamba-ssm, deep_ep, and deep_gemm). There are no code modifications, new features, breaking changes, or significant refactoring involved. The changes do not directly affect numerics, convergence, or performance of the codebase itself. Per the check criteria, test results documentation is not required for minor changes.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.

✨ Finishing touches

• 📝 Generate docstrings

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch tk/uv-tempermental

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 1

📜 Review details

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 855151b and 0c23e43.

⛔ Files ignored due to path filters (1)

• uv.lock is excluded by !**/*.lock

📒 Files selected for processing (1)

• pyproject.toml (2 hunks)

⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (3)

• GitHub Check: Lint check
• GitHub Check: Post submodule check comment / Comment on PR
• GitHub Check: Post automodel integration comment / Comment on PR

🔇 Additional comments (3)

pyproject.toml (3)

191-199: Reformatting no-build-isolation-package to trigger lock regeneration is sensible.

The multi-line format improves readability without functional changes to the dependency list. This pyproject.toml edit should indeed trigger uv.lock regeneration, which aligns with the PR goal to stabilize the lock.

---

1-307: Improve PR description and provide context for lock stabilization.

The PR description is incomplete and left mostly as template boilerplate. The commit message "re lock" is cryptic. While the actual code changes (adding version metadata for git-sourced dependencies and reformatting the no-build-isolation-package list) are reasonable steps to stabilize the lock, the PR lacks context on:

• What "experimental changes" caused lock instability?
• Why are these specific version pins the solution?
• Were there reproduction steps or failures in the lock file?
• What is the expected outcome?

Please update the PR description with clear context and consider linking any related issues.

---

224-250: No duplicate sections exist—AI summary is inaccurate, code is correct.

Verification confirms the pyproject.toml file contains exactly 5 [[tool.uv.dependency-metadata]] blocks (lines 224, 228, 234, 240, 246), one per package. There is no second "duplicated section." The AI summary's claim of "mirrored" metadata blocks under a duplicate section is incorrect. The code changes are correct; only the AI-generated description was misleading.