Azure Machine Learning for Visual Studio Code, previously called Visual Studio Code Tools for AI, is an extension to easily build, train, and deploy machine learning models to the cloud or the edge with Azure Machine Learning service.
Note: Public releases are always even numbered (e.g. 0.14.0). Nightly releases are odd numbered (0.13.0), which the build system handles automatically. When you release, make sure to bump the version to the next even number. Once that is merged, nightly releases will start releasing with the next odd number. Currently we are bumping the minor version for each release.
Checklist
Testing/endgame week
[x] Create a release branch from version of master intended to be released. Name it "Rel#.#.#", e : Rel0.14.0
[x] Update version in package.json in the release branch
[x] Update CHANGELOG.md with changes made in the release branch.
[x] Submit the above edits as a PR and have it reviewed.
[x] Run through the Testing Checklist below to find any potential regressions using the latest pre-release version published to the marketplace (using VS Code Insiders)
If hotfixes are required before release
[x] Make the hotfix change in master, then cherry-pick it to the release branch
[x] Test any affected areas from the Testing Checklist
Release week Target date for release should match VS Code, which is usually Wednesday
[x] Tag the commit in the release branch with the version number (e.g. v0.14.0). This will trigger a pipeline run in the release pipeline: https://dev.azure.com/monacotools/Monaco/_build?definitionId=320&_a=summary There is a manual review step, so this pipeline will build the release candidate, but won't release until you approve.
[x] Approve the Publish step in the release pipeline to allow the pipeline to upload/publish the vsix to the marketplace.
[x] Download the latest version from the marketplace (note, it may take up to 30 minutes and restarting vscode multiple times to force it to refresh the marketplace). Do a quick test to ensure the extension works as expected.
Testing Checklist
YAML Submissions
[x] Create workspace
[x] Create experiment
[x] Create environment
[x] Create aml cluster
[x] Create aml compute instance
[x] Dataset new version
[x] Local experiment submission
[x] Remote experiment submission
[x] Reference dataset during run submission
[x] Update environment (e.g. changing dependencies updates the version)
[x] Jupyter extension integration ("Pick how to connect to Jupyter" list and recent connection in that list)
[No regression] After selecting remote server there is no reload prompt or a prompt to select a new kernel (keeps executing locally). Need to manually select the appropriate remote kernel.
[x] Delete workspace
Files
[x] Test Azure ML snippets on Python files (Ctrl + Space -> import-basic)
[x] Test YAML language support
[x] Test YAML template generation
Remote experience
[x] Test non-standard management endpoints (e.g. gov cloud) @shsuman
Note: Public releases are always even numbered (e.g. 0.14.0). Nightly releases are odd numbered (0.13.0), which the build system handles automatically. When you release, make sure to bump the version to the next even number. Once that is merged, nightly releases will start releasing with the next odd number. Currently we are bumping the minor version for each release.
Checklist
Testing/endgame week
If hotfixes are required before release
Release week Target date for release should match VS Code, which is usually Wednesday
Publishstep in the release pipeline to allow the pipeline to upload/publish the vsix to the marketplace.Testing Checklist
YAML Submissions
Tree operations
Files
Remote experience