Fetch the latest micro version of python from conda in cross version …

…tests to avoid `ResolvePackageNotFound` error (mlflow#4559) * use conda search to obtain python version Signed-off-by: harupy <17039389+harupy@users.noreply.github.com> * temp change Signed-off-by: harupy <17039389+harupy@users.noreply.github.com> * debug Signed-off-by: harupy <17039389+harupy@users.noreply.github.com> * double quotes Signed-off-by: harupy <17039389+harupy@users.noreply.github.com> * Use fixed strings Signed-off-by: harupy <17039389+harupy@users.noreply.github.com> * comment Signed-off-by: harupy <17039389+harupy@users.noreply.github.com>
BLACKBUCK-LABS · Jul 14, 2021 · 8ec8603 · 8ec8603
1 parent 644d11a
commit 8ec8603
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 1 deletion.
diff --git a/.github/workflows/cross-version-tests.yml b/.github/workflows/cross-version-tests.yml
@@ -78,7 +78,10 @@ jobs:
           else
             python_version=3.6
           fi
-          echo "::set-output name=python-version::$python_version"
+          # Use a python version that exists on Anaconda to ensure `mlflow serve` can successfully
+          # create a conda environment for serving an MLflow model.
+          latest_micro_version=$($CONDA/bin/conda search python | grep -F $python_version | tail -1 | tr -s ' ' | cut -d' ' -f2)
+          echo "::set-output name=python-version::$latest_micro_version"
       - uses: actions/setup-python@v2
         with:
           python-version: ${{ steps.get-python-version.outputs.python-version }}

diff --git a/mlflow/onnx.py b/mlflow/onnx.py
@@ -28,6 +28,7 @@
 from mlflow.tracking._model_registry import DEFAULT_AWAIT_MAX_SLEEP_SECONDS
 
 FLAVOR_NAME = "onnx"
+# TEMPORARY CHANGE to trigger cross version tests
 
 
 @experimental