Repository navigation
style: reorder one re-export #1648 left unsorted on main - #1651
Merged
Merged
Conversation
`cargo fmt --all -- --check` fails on an unmodified `origin/main` (90ddd28): crates/onnx-runtime-session/src/lib.rs:38 `pub use onnx_runtime_ep_api::DeviceGraphSlot;` was added above the existing `WorkspaceRequirement` re-export instead of in sorted order. One-line swap, pure `cargo fmt --all` output. Formatting is a required check, so this blocks every open PR. This is the fourth main-is-red repair today (#1640, #1642, #1649, this one). The pattern is consistent: quality gates are being evaluated on PR branches before merge but not on the merge result, so each merge can and does land violations that then fail whoever opens the next PR. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #1651 +/- ##
==========================================
+ Coverage 80.84% 81.64% +0.80%
==========================================
Files 383 382 -1
Lines 175886 177266 +1380
Branches 175886 177266 +1380
==========================================
+ Hits 142199 144735 +2536
+ Misses 28786 27596 -1190
- Partials 4901 4935 +34
Flags with carried forward coverage won't be shown. Click here to find out more.
🚀 New features to boost your workflow:
|
This was referenced Aug 21, 2026
justinchuby
added a commit
that referenced
this pull request
Aug 21, 2026
`Rust quality → Check formatting` is failing on `main` at `6923a016b` (#1652). That job runs its steps sequentially, so while formatting is red **every other check in it is skipped** — clippy, publish order, the dispatch-manifest lints, feature-gate coverage, all of it. Every open PR is blocked and none of them are getting linted. Three sites, all in `crates/onnx-genai-engine/src/native_decode/cuda.rs`: two `verify_graph_phase` assignments (`:1815`, `:1823`) and one `assert_eq!` in `verify_capture_helper_tests` (`:7042`). Straight `cargo fmt --all` output, no hand edits. **Verified inert.** The before/after texts are identical after stripping whitespace *and* trailing commas — the only non-whitespace delta is commas rustfmt adds before a closing delimiter when it breaks a call across lines, which are semantically meaningless in Rust. Reproduced on a pristine `origin/main` worktree first, so this is main's breakage and not an artifact of my branch. ### This is the fifth time today `main` has been red on formatting or clippy five separate times in one day: #1637/#1641 (fixed by #1640), #1644 (#1642), #1648 (#1649), #1647/#1648 (#1651), and now #1652. The cause is structural, not carelessness. Required checks run on a PR's **merge ref**, but nothing re-runs them on `main` **after** the merge, so two PRs that are each green against an older base can land in sequence and leave the result red. Because the quality job is sequential, the breakage also masks every later step in it. The cost lands on whoever opens the next PR, who then has to distinguish "my change broke this" from "main was already broken" — a full CI round-trip each time. Two things would fix it, either one sufficient: - a **merge queue**, which tests the actual post-merge result; or - running **`Rust quality` on `main` post-merge**, which at least detects it immediately and attributes it correctly. I have now spent four PRs on this. I would rather not spend a fifth. cc @justinchuby Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
cargo fmt --all -- --checkfails on an unmodifiedorigin/main(90ddd284e):pub use onnx_runtime_ep_api::DeviceGraphSlot;was added above the existingWorkspaceRequirementre-export rather than in sorted order. One-line swap, purecargo fmt --alloutput.Formatting is a required check, so this blocks every open PR regardless of contents. It surfaced on an unrelated CPU-kernel PR (#1628).
This is the fourth main-is-red repair today
dead_code,native_decode/cuda.rsThe pattern is consistent and worth fixing at the source: quality gates run on PR branches before merge but not on the merge result, so any merge can land violations that then fail whoever opens the next PR. Because the gates are sequential — clippy steps only run once formatting passes — each breakage costs a full CI round-trip to even discover, and they arrive one at a time.
Two concrete options: enable a merge queue (gates run on the merge result), or run the
Rust qualityjob onmainpost-merge so the break is attributed to the PR that caused it instead of the next unrelated one.Also still red on main and not fixed here, because I cannot reproduce it locally and it is not mine:
Rust (Windows ARM64)→ Test cross-platform offline crates has been failing on main since at least73e6fe15a(it is non-required, so it does not block merges).