microsoft
Mangler enabled in hyenie/CI part of pipeline?
I noticed recently that PR's takle longer to merge because the mangler is enabled in the compile/hygenie step, e.g. https://github.com/microsoft/vscode/actions/runs/19175726948/job/54819910297?pr=276137
I'm guessing the intention is to catch failures that may happen due to private access in mangling. However, this change won't actually catch those -- those failures happen when running tests in code that has already been mangled. That doesn't happen in this CI check, as-is the CI is only checking that the mangler itself doesn't error out, which I don't think has ever been the issue.
We'd need to actually wait and run tests on mangled code, but would make PR merging even slower...
Great callout. The intention is to close the gap between what is being run as part of the PR build vs. what is being run in the CI/product build. Anything that we can do to narrow the gap saves people precious time when the CI/product build turns red.
