Centralized metric and specification definitions for the LSST Science Pipeline Verification Framework.

For background on the lsst.verify framework, see the SQR-017 (design) and SQR-019 (demo) technotes.

This verify_metrics package is a dependency of the verify package, and available through the lsst_distrib top-level package. See https://pipelines.lsst.io for details on installing lsst_distrib.

Metrics are measureable things. They can be astronomy concepts (an astrometric RMS), or computational (a task's timing or memory usage).

Metrics that are measured by packages are registered here in verify_metrics by adding or extending a metric definition YAML file in the metrics/ directory. See metrics/README.rst for more information.

Specifications are tests of metric measurements. A specification might say that a metric measurement should be less than a certain value if that measurement was made under a certain set of conditions (provenance). Whenever metrics are measured, DM's SQUASH harness tests those measurements against specifications. Depending on the pass/fail outcome, subscribers can receive notifications (alerts).

New specifications are registered here in verify_metrics by adding or extending a specification definition YAML file in the specs/ directory. See specs/README.rst for more information.

If you're not part of the LSST Project, please post your question or issue in our support forum. It's easy to create a Community forum account.

We don't use GitHub Issues.

If you're part of the LSST Project, please create a JIRA ticket. Use the Verification component component and verify_metrics label.

• This repository should only contain metric and specification YAML files. Don't add Python code; instead consider extending the lsst.verify API in the verify package.
• Follow LSST Data Management's workflow for contributions.

This product includes software developed by the LSST Project. See the COPYRIGHT file.

This product's source code is licensed under the terms of GPLv3 (see LICENSE), and all documentation content is licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.