Switch CI from bors to merge queue

.github/workflows/clippy_dev.yml

@@ -1,10 +1,7 @@
 name: Clippy Dev Test
 
 on:
-  push:
-    branches:
-      - auto
-      - try
+  merge_group:
   pull_request:
     # Only run on paths, that get checked by the clippy_dev tool
     paths:
@@ -47,28 +44,21 @@ jobs:
         cargo check
         git reset --hard HEAD
 
-  # These jobs doesn't actually test anything, but they're only used to tell
-  # bors the build completed, as there is no practical way to detect when a
-  # workflow is successful listening to webhooks only.
-  #
-  # ALL THE PREVIOUS JOBS NEED TO BE ADDED TO THE `needs` SECTION OF THIS JOB!
-
-  end-success:
-    name: bors dev test finished
-    if: github.event.pusher.name == 'bors' && success()
-    runs-on: ubuntu-latest
-    needs: [clippy_dev]
-
-    steps:
-      - name: Mark the job as successful
-        run: exit 0
-
-  end-failure:
-    name: bors dev test finished
-    if: github.event.pusher.name == 'bors' && (failure() || cancelled())
+  conclusion_dev:
+    needs: [ clippy_dev ]
+    # We need to ensure this job does *not* get skipped if its dependencies fail,
+    # because a skipped job is considered a success by GitHub. So we have to
+    # overwrite `if:`. We use `!cancelled()` to ensure the job does still not get run
+    # when the workflow is canceled manually.
+    #
+    # ALL THE PREVIOUS JOBS NEED TO BE ADDED TO THE `needs` SECTION OF THIS JOB!
+    if: ${{ !cancelled() }}
     runs-on: ubuntu-latest
-    needs: [clippy_dev]
-
     steps:
-      - name: Mark the job as a failure
-        run: exit 1
+      # Manually check the status of all dependencies. `if: failure()` does not work.
+      - name: Conclusion
+        run: |
+          # Print the dependent jobs to see them in the CI log
+          jq -C <<< '${{ toJson(needs) }}'
+          # Check if all jobs that we depend on (in the needs array) were successful.
+          jq --exit-status 'all(.result == "success")' <<< '${{ toJson(needs) }}'

.github/workflows/clippy_bors.yml → .github/workflows/clippy_mq.yml

@@ -1,10 +1,7 @@
-name: Clippy Test (bors)
+name: Clippy Test (merge queue)
 
 on:
-  push:
-    branches:
-      - auto
-      - try
+  merge_group:
 
 env:
   RUST_BACKTRACE: 1
@@ -13,11 +10,6 @@ env:
   CARGO_INCREMENTAL: 0
   RUSTFLAGS: -D warnings
 
-concurrency:
-  # For a given workflow, if we push to the same branch, cancel all previous builds on that branch.
-  group: "${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}"
-  cancel-in-progress: true
-
 defaults:
   run:
     shell: bash
@@ -218,28 +210,21 @@ jobs:
       env:
         INTEGRATION: ${{ matrix.integration }}
 
-  # These jobs doesn't actually test anything, but they're only used to tell
-  # bors the build completed, as there is no practical way to detect when a
-  # workflow is successful listening to webhooks only.
-  #
-  # ALL THE PREVIOUS JOBS NEED TO BE ADDED TO THE `needs` SECTION OF THIS JOB!
-
-  end-success:
-    name: bors test finished
-    if: github.event.pusher.name == 'bors' && success()
+  conclusion:
+    needs: [ changelog, base, metadata_collection, integration_build, integration ]
+    # We need to ensure this job does *not* get skipped if its dependencies fail,
+    # because a skipped job is considered a success by GitHub. So we have to
+    # overwrite `if:`. We use `!cancelled()` to ensure the job does still not get run
+    # when the workflow is canceled manually.
+    #
+    # ALL THE PREVIOUS JOBS NEED TO BE ADDED TO THE `needs` SECTION OF THIS JOB!
+    if: ${{ !cancelled() }}
     runs-on: ubuntu-latest
-    needs: [changelog, base, metadata_collection, integration_build, integration]
-
-    steps:
-      - name: Mark the job as successful
-        run: exit 0
-
-  end-failure:
-    name: bors test finished
-    if: github.event.pusher.name == 'bors' && (failure() || cancelled())
-    runs-on: ubuntu-latest
-    needs: [changelog, base, metadata_collection, integration_build, integration]
-
     steps:
-      - name: Mark the job as a failure
-        run: exit 1
+      # Manually check the status of all dependencies. `if: failure()` does not work.
+      - name: Conclusion
+        run: |
+          # Print the dependent jobs to see them in the CI log
+          jq -C <<< '${{ toJson(needs) }}'
+          # Check if all jobs that we depend on (in the needs array) were successful.
+          jq --exit-status 'all(.result == "success")' <<< '${{ toJson(needs) }}'

.github/workflows/clippy.yml → .github/workflows/clippy_pr.yml

@@ -1,17 +1,6 @@
 name: Clippy Test
 
 on:
-  push:
-    # Ignore bors branches, since they are covered by `clippy_bors.yml`
-    branches-ignore:
-      - auto
-      - try
-    # Don't run Clippy tests, when only text files were modified
-    paths-ignore:
-    - 'COPYRIGHT'
-    - 'LICENSE-*'
-    - '**.md'
-    - '**.txt'
   pull_request:
     # Don't run Clippy tests, when only text files were modified
     paths-ignore:
@@ -35,7 +24,7 @@ concurrency:
 
 jobs:
   base:
-    # NOTE: If you modify this job, make sure you copy the changes to clippy_bors.yml
+    # NOTE: If you modify this job, make sure you copy the changes to clippy_mq.yml
     runs-on: ubuntu-latest
 
     steps:

.github/workflows/remark.yml

@@ -1,10 +1,7 @@
 name: Remark
 
 on:
-  push:
-    branches:
-      - auto
-      - try
+  merge_group:
   pull_request:
     paths:
     - '**.md'
@@ -45,28 +42,21 @@ jobs:
     - name: Build mdbook
       run: mdbook build book
 
-  # These jobs doesn't actually test anything, but they're only used to tell
-  # bors the build completed, as there is no practical way to detect when a
-  # workflow is successful listening to webhooks only.
-  #
-  # ALL THE PREVIOUS JOBS NEED TO BE ADDED TO THE `needs` SECTION OF THIS JOB!
-
-  end-success:
-    name: bors remark test finished
-    if: github.event.pusher.name == 'bors' && success()
-    runs-on: ubuntu-latest
-    needs: [remark]
-
-    steps:
-      - name: Mark the job as successful
-        run: exit 0
-
-  end-failure:
-    name: bors remark test finished
-    if: github.event.pusher.name == 'bors' && (failure() || cancelled())
+  conclusion_remark:
+    needs: [ remark ]
+    # We need to ensure this job does *not* get skipped if its dependencies fail,
+    # because a skipped job is considered a success by GitHub. So we have to
+    # overwrite `if:`. We use `!cancelled()` to ensure the job does still not get run
+    # when the workflow is canceled manually.
+    #
+    # ALL THE PREVIOUS JOBS NEED TO BE ADDED TO THE `needs` SECTION OF THIS JOB!
+    if: ${{ !cancelled() }}
     runs-on: ubuntu-latest
-    needs: [remark]
-
     steps:
-      - name: Mark the job as a failure
-        run: exit 1
+      # Manually check the status of all dependencies. `if: failure()` does not work.
+      - name: Conclusion
+        run: |
+          # Print the dependent jobs to see them in the CI log
+          jq -C <<< '${{ toJson(needs) }}'
+          # Check if all jobs that we depend on (in the needs array) were successful.
+          jq --exit-status 'all(.result == "success")' <<< '${{ toJson(needs) }}'

CONTRIBUTING.md

@@ -21,7 +21,6 @@ All contributors are expected to follow the [Rust Code of Conduct].
     - [Rust Analyzer](#rust-analyzer)
   - [How Clippy works](#how-clippy-works)
   - [Issue and PR triage](#issue-and-pr-triage)
-  - [Bors and Homu](#bors-and-homu)
   - [Contributions](#contributions)
   - [License](#license)
 
@@ -213,26 +212,13 @@ We have prioritization labels and a sync-blocker label, which are described belo
 Or rather: before the sync this should be addressed,
 e.g. by removing a lint again, so it doesn't hit beta/stable.
 
-## Bors and Homu
-
-We use a bot powered by [Homu][homu] to help automate testing and landing of pull
-requests in Clippy. The bot's username is @bors.
-
-You can find the Clippy bors queue [here][homu_queue].
-
-If you have @bors permissions, you can find an overview of the available
-commands [here][homu_instructions].
-
 [triage]: https://forge.rust-lang.org/release/triage-procedure.html
 [l-crash]: https://github.com/rust-lang/rust-clippy/labels/L-crash
 [l-bug]: https://github.com/rust-lang/rust-clippy/labels/L-bug
 [p-low]: https://github.com/rust-lang/rust-clippy/labels/P-low
 [p-medium]: https://github.com/rust-lang/rust-clippy/labels/P-medium
 [p-high]: https://github.com/rust-lang/rust-clippy/labels/P-high
 [l-sync-blocker]: https://github.com/rust-lang/rust-clippy/labels/L-sync-blocker
-[homu]: https://github.com/rust-lang/homu
-[homu_instructions]: https://bors.rust-lang.org/
-[homu_queue]: https://bors.rust-lang.org/queue/clippy
 
 ## Contributions
 
@@ -244,7 +230,7 @@ All PRs should include a `changelog` entry with a short comment explaining the c
 "what do you believe is important from an outsider's perspective?" Often, PRs are only related to a single property of a
 lint, and then it's good to mention that one. Otherwise, it's better to include too much detail than too little.
 
-Clippy's [changelog] is created from these comments. Every release, someone gets all commits from bors with a
+Clippy's [changelog] is created from these comments. Every release, someone gets all merge commits with a
 `changelog: XYZ` entry and combines them into the changelog. This is a manual process.
 
 Examples:

book/src/development/README.md

@@ -53,7 +53,6 @@ book](../lints.md).
 > - IDE setup
 > - High level overview on how Clippy works
 > - Triage procedure
-> - Bors and Homu
 
 [ast]: https://rustc-dev-guide.rust-lang.org/syntax-intro.html
 [hir]: https://rustc-dev-guide.rust-lang.org/hir.html