WIP: {platform}-github tags

.github/workflows/ci-matrix-gen.yml

@@ -12,6 +12,14 @@ on:
       jobs_to_run:
         required: true
         type: string
+      logs_retention_days:
+        required: false
+        type: number
+        default: 15
+      artifacts_retention_days:
+        required: false
+        type: number
+        default: 7
 
 jobs:
   generate-tier1:
@@ -55,7 +63,6 @@ jobs:
   tier1:
     needs: generate-tier1
     runs-on: ${{ matrix.os }}
-    #env: ${{ matrix.env }}
     strategy:
       fail-fast: false
       matrix:
@@ -150,6 +157,7 @@ jobs:
         if: ${{ matrix.mx_version }}
         shell: bash
         run: |
+          git config --global http.timeout 600
           git clone https://github.com/graalvm/mx
           if [[ "${RUNNER_OS}" == "Windows" ]]; then
             ./mx/mx.cmd fetch-jdk -A --jdk-id labsjdk-ce-latest
@@ -255,22 +263,23 @@ jobs:
         with:
           name: ${{ matrix.provide_artifact[0] }}
           path: main/${{ matrix.provide_artifact[0] }}.tar
-          retention-days: 1
+          retention-days: ${{ inputs.artifacts_retention_days || 7 }}
 
       - name: Upload logs
         if: ${{ matrix.logs }}
         uses: actions/upload-artifact@v5
+        continue-on-error: true
         with:
           name: ${{ format('{0}_logs', matrix.name) }}
           path: |
             ${{ matrix.logs }}
-          retention-days: 15
+          retention-days: ${{ inputs.logs_retention_days || 15 }}
+          if-no-files-found: ignore
 
   tier2:
     if: ${{ success() || inputs.jobs_to_run }}
     needs: [generate-tier2, tier1]
     runs-on: ${{ matrix.os }}
-    #env: ${{ matrix.env }}
     strategy:
       fail-fast: false
       matrix:
@@ -281,7 +290,6 @@ jobs:
     if: ${{ success() || inputs.jobs_to_run }}
     needs: [generate-tier3, tier2]
     runs-on: ${{ matrix.os }}
-    #env: ${{ matrix.env }}
     strategy:
       fail-fast: false
       matrix:

.github/workflows/ci-unittest-retagger.yml

@@ -15,6 +15,7 @@ jobs:
     uses: ./.github/workflows/ci-matrix-gen.yml
     with:
       jobs_to_run: ^(?=.*python-svm-build).*$
+      artifacts_retention_days: 0
 
   run-retagger:
     if: success()
@@ -27,9 +28,12 @@ jobs:
     name: Run retagger on ${{ matrix.target }}
     with:
       jobs_to_run: ${{ github.event.inputs.jobs_to_run || format('^python-unittest-retagger.*{0}.*', matrix.target) }}
+      logs_retention_days: 7
+      artifacts_retention_days: 0
 
   merge_all_reports:
     runs-on: ubuntu-latest
+    if: always()
     needs: run-retagger
     env:
       GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
@@ -55,7 +59,7 @@ jobs:
           git config --global user.name "Retagger Workflow" 
           git config --global user.email "Retagger_Workflow@oracle.com"
           git fetch origin
-          git checkout -b weekly_retagger_${{ github.run_id }} origin/master
+          git checkout -b weekly_retagger_${{ github.run_id }} origin/${{ github.head_ref || github.ref_name }} 
 
       - name: Merge retagger reports
         env:
@@ -69,14 +73,14 @@ jobs:
           for os in "${os_list[@]}"; do
             echo "Merging tags for $os"
             python3 .github/scripts/merge_retagger_results.py --dir ../retagger-reports --outfile "../retagger-reports/reports-merged-$os.json" --pattern "*$os*" || true
-            python3 graalpython/com.oracle.graal.python.test/src/runner.py merge-tags-from-report "../retagger-reports/reports-merged-$os.json" --platform "$os" || true
+            python3 graalpython/com.oracle.graal.python.test/src/runner.py merge-tags-from-report "../retagger-reports/reports-merged-$os.json" --platform "$os-github" || true
             git add -A
             git commit -m "Apply retags for $os" || true
           done
 
       - name: Push updates to a new PR
         working-directory: main
         run: |
-          [ "$(git diff origin/master...HEAD | wc -l)" -eq 0 ] && echo 'Nothing to push' || git push --set-upstream origin weekly_retagger_${{ github.run_id }}
+          [ "$(git diff origin/${{ github.head_ref || github.ref_name }}...HEAD | wc -l)" -eq 0 ] && echo 'Nothing to push' || git push --set-upstream origin weekly_retagger_${{ github.run_id }}
           gh label create "weekly-retagger" --color "BDDFDF" -d "[DO NOT ADD] Automatically set to weekly retagger PR" || echo "Label 'weekly-retagger' already exists"
           gh pr create -B master --title "[WORKFLOW] Weekly Retagger: Update tags" --body "Applied weekly retags on $os_list" -l "weekly-retagger" || echo "Nothing to push"

.github/workflows/ci-unittests.yml

@@ -11,6 +11,8 @@ jobs:
     uses: ./.github/workflows/ci-matrix-gen.yml
     with:
       jobs_to_run: ^(?=.*python-svm-build).*$
+      logs_retention_days: 0
+      artifacts_retention_days: 0
 
   run-tests:
     if: github.event.pull_request.draft == false && success()

ci.jsonnet

@@ -101,7 +101,7 @@
     local GRAAL_JDK_LATEST          = "graal-jdk-latest",
     local TAGGED_UNITTESTS_SPLIT    = 8,
     local COVERAGE_SPLIT            = 3,
-    local RETAGGER_SPLIT            = 12,
+    local RETAGGER_SPLIT            = 16,
 
     // -----------------------------------------------------------------------------------------------------------------
     // gates
@@ -242,10 +242,10 @@
             "linux:amd64:jdk-latest"     : tier1,
         }),
         "python-unittest-retagger": ut_retagger + platform_spec(no_jobs) + batches(RETAGGER_SPLIT) + platform_spec({
-            "linux:amd64:jdk-latest"     : weekly    + t("20:00:00"),
-            "linux:aarch64:jdk-latest"   : weekly    + t("20:00:00"),
-            "darwin:aarch64:jdk-latest"  : weekly    + t("20:00:00"),
-            "windows:amd64:jdk-latest"   : weekly    + t("20:00:00"),
+            "linux:amd64:jdk-latest"     : weekly    + t("20:00:00") + require(GPY_NATIVE_STANDALONE),
+            "linux:aarch64:jdk-latest"   : weekly    + t("20:00:00") + require(GPY_NATIVE_STANDALONE),
+            "darwin:aarch64:jdk-latest"  : weekly    + t("20:00:00") + require(GPY_NATIVE_STANDALONE),
+            "windows:amd64:jdk-latest"   : weekly    + t("20:00:00") + require(GPY_NATIVE_STANDALONE),
         }),
         "python-coverage-jacoco-tagged": cov_jacoco_tagged + batches(COVERAGE_SPLIT) + platform_spec(no_jobs) + platform_spec({
             "linux:amd64:jdk21"          : weekly    + t("20:00:00"),

ci/python-gate.libsonnet

@@ -151,14 +151,14 @@
         darwin: {
             common: ENV_POSIX + {
                 LC_CTYPE: "en_US.UTF-8",
-                PATH: utils.path_combine(ENVIRONMENT.common.PATH, "$PATH:$PYTHON3_HOME:$MUSL_TOOLCHAIN/bin"),
+                PATH: utils.path_combine(ENVIRONMENT.common.PATH, "$PYTHON3_HOME:$PATH:$MUSL_TOOLCHAIN/bin"),
             },
             amd64: {},
             aarch64: {},
         },
         windows: {
             common: {
-                PATH: "%MAVEN_HOME%;%PATH%",
+                PATH: "%MAVEN_HOME%\\bin;%PATH%",
             },
             amd64: {},
             aarch64: {},
@@ -509,7 +509,7 @@
                 "mx", "graalpytest", "--svm",
                 "--tagged", "--all", "--continue-on-collection-errors", ".",
                 # More workers doesn't help, the job is bottlenecked on all the timeouts in test_asyncio
-                "-n", "6",
+                "-n", "4",
                 # The default timeout is very generous to allow for infrastructure flakiness,
                 # but we don't want to auto tag tests that take a long time
                 "--timeout-factor", "0.3",

graalpython/com.oracle.graal.python.test/src/runner.py

@@ -85,6 +85,9 @@
 DISABLE_JIT_ENV = {'GRAAL_PYTHON_VM_ARGS': '--experimental-options --engine.Compilation=false'}
 
 GITHUB_CI = os.environ.get("GITHUB_CI", None)
+if GITHUB_CI:
+    PLATFORM_KEYS.add("github")
+    CURRENT_PLATFORM += "-github"
 
 # We leave the JIT enabled for the tests themselves, but disable it for subprocesses
 # noinspection PyUnresolvedReferences
@@ -907,7 +910,13 @@ def run_in_subprocess_and_watch(self):
 
 
 def platform_keys_match(items: typing.Iterable[str]):
-    return any(all(key in PLATFORM_KEYS for key in item.split('-')) for item in items)
+    matches = []
+    for item in items:
+        if GITHUB_CI:
+            if not "github" in item.split('-'):
+                continue
+        matches.append(all(key in PLATFORM_KEYS for key in item.split('-')))
+    return any(matches)
 
 
 @dataclass
@@ -1132,7 +1141,7 @@ def collect_module(test_file: TestFile, specifiers: list[TestSpecifier], use_tag
         loader = TopLevelFunctionLoader() if config.run_top_level_functions else unittest.TestLoader()
         tagged_ids = None
         if use_tags and config.tags_dir:
-            tagged_ids = [tag.test_id for tag in read_tags(test_file) if platform_keys_match(tag.keys) and not tag.is_platform_excluded(CURRENT_PLATFORM)]
+            tagged_ids = [tag.test_id for tag in read_tags(test_file) if platform_keys_match(tag.keys)]
             if not tagged_ids:
                 return None
         test_module = test_path_to_module(test_file)
@@ -1208,52 +1217,36 @@ def collect(all_specifiers: list[TestSpecifier], *, use_tags=False, ignore=None,
 class Tag:
     test_id: TestId
     keys: frozenset[str]
-    excluded_keys: frozenset[str]
     is_exclusion: bool
     comment: str | None = False
 
     @classmethod
     def for_key(cls, test_id, key) -> 'Tag':
-        return Tag(test_id, frozenset({key}), excluded_keys=frozenset(), is_exclusion=False)
+        return Tag(test_id, frozenset({key}), is_exclusion=False)
 
     def with_key(self, key: str) -> 'Tag':
-        if GITHUB_CI:
-            return Tag(self.test_id, self.keys, self.excluded_keys - {key}, is_exclusion=self.is_exclusion)
-
-        return Tag(self.test_id, self.keys | {key}, self.excluded_keys, is_exclusion=self.is_exclusion)
+        return Tag(self.test_id, self.keys | {key}, is_exclusion=self.is_exclusion)
 
     def without_key(self, key: str) -> 'Tag | None':
         return self.without_keys({key})
 
     def without_keys(self, keys: set[str]) -> 'Tag | None':
-        # disable test/platform only for github, not completely for internal ci
-        if GITHUB_CI:
-            exclusions = frozenset(list(self.excluded_keys) + list(keys))
-            return Tag(self.test_id, self.keys, exclusions, is_exclusion=self.is_exclusion)
-
         keys = self.keys - keys
         if keys:
             if keys == self.keys:
                 return self
-            return Tag(self.test_id, keys, self.excluded_keys, is_exclusion=self.is_exclusion)
+            return Tag(self.test_id, keys, is_exclusion=self.is_exclusion)
 
     def is_platform_excluded(self, key: str) -> bool:
-        return key in self.excluded_keys
+        return key not in self.keys
 
-    def get_keys_as_str(self) -> list[str]:
-        keys = []
-        for key in sorted(self.keys):
-            if "$" in key: print(f"[WARNING]: Invalid key with $ found in tag keys: {key}, {self.keys}, {self.excluded_keys}")
-            keys.append(key if key not in self.excluded_keys else f"${key}")
-        return keys
-
     def __str__(self):
         s = ''
         if self.is_exclusion:
             s += '!'
         s += self.test_id.test_name
         if self.keys:
-            s += f' @ {",".join(self.get_keys_as_str())}'
+            s += f' @ {",".join(sorted(self.keys))}'
         if self.comment:
             s = f'{self.comment}{s}'
         return s
@@ -1282,19 +1275,13 @@ def read_tags(test_file: TestFile, allow_exclusions=False) -> list[Tag]:
                     is_exclusion = True
                     test = test.removeprefix('!')
 
+
                 if not keys and not is_exclusion:
                     log(f'WARNING: invalid tag {test}: missing platform keys')
 
-                keys = keys.split(',') if keys else []
-                excluded_keys = []
-                for key in keys:
-                    if key.startswith('$'):
-                        excluded_keys.append(key.removeprefix('$'))
-
                 tag = Tag(
                     TestId(test_path, test),
-                    frozenset(keys),
-                    excluded_keys=frozenset(excluded_keys),
+                    frozenset(keys.split(',')) if keys else frozenset(keys),
                     is_exclusion=is_exclusion,
                     comment=comment,
                 )

graalpython/com.oracle.graal.python.test/src/tests/test_wheel.py

@@ -48,7 +48,8 @@
 
 from tests.testlib_helper import build_testlib
 
-
+# cannot be moved to tagged as it uses testlib_helper
+@unittest.skipIf(os.environ.get("GITHUB_CI"), "Skip on Github CI")
 class TestWheelBuildAndRun(unittest.TestCase):
     def test_build_install_and_run(self):
         # Build a C library and a wheel that depends on it. Then run it through repair_wheel to vendor the library in and verify that it can be imported