mudler / mudler/vllm.cpp

documentation-checkpoint reds on main: four recent commits reached main without arriving on a task branch

Open
#2,375 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
C++
Stars
423
Forks
53
Avg merge
20h 26m
Merged PRs (30d)
310

Description

Observed

The documentation-checkpoint CI job fails on main's head 8846e3c4b (job 99351159519) with four attribution errors. The gate runs on main pushes only, which is why no open PR surfaces it:

ERROR: e1b5df1a6: repository change (tests/vllm/multimodal/ltx2_video_fixture.h) reached main without arriving on a task branch.
ERROR: c2501cc71: repository change (.w9a/bridge2.txt, .w9a/build.pid, .w9a/final_moe.txt, .w9a/green.txt, ... (+19)) reached main without arriving on a task branch.
ERROR: 330599b26: repository change (tests/vllm/models/test_glm5_next_forward.cpp) reached main without arriving on a task branch.
ERROR: 8846e3c4b: repository change (src/vt/cuda/cuda_quant_dot.cu) reached main without arriving on a task branch.

What is unclear, and what the issue asks

Whether these are (a) genuine direct-to-main landings, or (b) false positives of the branch-correlation heuristic — squash-merged PRs lose their branch name at the forge, and a branch named anything other than row/<ID> (for example the docs/* branches visible in the ref listing) may be invisible to the correlation. The c2501cc71 case is suspicious in both directions: its message says it DROPS W9a's scratch directory, yet the gate attributes the .w9a/* changes to it, which implies the scratch files were first committed to main by some earlier commit the gate did not flag.

Owed

The gate's owner reconciles the four commits against their actual landing paths (forge PR numbers exist for squash merges) and either repairs the process that landed them or repairs the correlation so a squash-merged row/<ID> PR is recognized. Likely owning rows by content: MODEL-MM-GLM53-FLASH (.w9a/, test_glm5_next_forward.cpp), QUANT-CUDA-IQ4XS-IQ2XS (cuda_quant_dot.cu), and the LTX2 fixture's row for e1b5df1a6. Until reconciled, documentation-checkpoint stays red on every main push and masks the next genuine violation.

Context found while classifying CI for #2368/#2370/#2373; none of those PRs touch this gate.

FOLLOWING_AGENTS_PROTOCOL

Following-Agents-Protocol: true
AI-Assisted: true
Assisted-by: AGENT:zai-glm-5.3-flash [maki]

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with the documentation-checkpoint job 99351159519 on main head 8846e3c4b and inspect the four reported commits: e1b5df1a6, c2501cc71, 330599b26, and 8846e3c4b. Reconcile each with its forge PR and landing path, then verify that the gate correctly recognizes valid squash-merged task branches and remains green without masking genuine violations.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp, github-actions
Domain
ci-cd, devops
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Active
Clarity
Mostly clear
Newbie friendliness
42/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.