Sometimes (intermittently?) codecov has no data for either the base or head commit of a PR and can produce no coverage statistics. We don't want the GitHub checks to fail for spurious operational reasons like this, so mark these coverage checks as informational only.
While it's good to pay attention to the coverage reports, I don't think we veto PRs on the basis of degraded coverage. So at least while there are intermittent spurious failures, we don't lose anything by stopping codecov from putting an :x: on builds.
Sometimes (intermittently?) codecov has no data for either the base or head commit of a PR and can produce no coverage statistics. We don't want the GitHub checks to fail for spurious operational reasons like this, so mark these coverage checks as informational only.
For example here the base commit previously failed CI tests, and at several commits in #875 the HEAD statistics were missing so codecov/patch spuriously failed the build checks.
While it's good to pay attention to the coverage reports, I don't think we veto PRs on the basis of degraded coverage. So at least while there are intermittent spurious failures, we don't lose anything by stopping codecov from putting an :x: on builds.