Closed KSXGitHub closed 10 months ago
group main pr
----- ---- --
tarball/download_dependency 1.00 6.7±0.20ms 644.2 KB/sec 1.03 7.0±1.73ms 622.9 KB/sec
All modified and coverable lines are covered by tests :white_check_mark:
Comparison is base (
6b00165
) 87.11% compared to head (d031d26
) 87.04%. Report is 1 commits behind head on main.:exclamation: Current head d031d26 differs from pull request most recent head 0e8bfe6. Consider uploading reports for the commit 0e8bfe6 to get more accurate results
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
Command | Mean [ms] | Min [ms] | Max [ms] | Relative |
---|---|---|---|---|
pacquet@HEAD |
138.4 ± 8.4 | 123.3 | 147.8 | 1.05 ± 0.09 |
pacquet@main |
132.3 ± 7.2 | 123.4 | 145.5 | 1.00 |
How does this mitigate the issue? The snapshot will be invalid if any of the files from the package will change. Or a package will get a new file.
How does this mitigate the issue?
Create an actual mock registry that we have full control over. Or maybe just a verdaccio proxy is sufficient?
The snapshot will be invalid if any of the files from the package will change.
Right now, the package versions are pinned. This does mean that we cannot test something involving version ranges (such as pacquet add foo@latest
), so eventually, we would have to properly solve the problem.
For pnpm we use a verdaccio proxy with prepublished packages for testing: https://github.com/pnpm/registry-mock
I think we should use the same in this repo.
Changes in the npm registry cause unnecessary snapshot failures in our tests. The ideal solution is to implement a fake registry that we can control, but until then, filtering them out is good enough.