search

npms-io / npms-analyzer

The analyzer behind https://npms.io
MIT License
314 stars 37 forks source link

Frequency tests should consider current date #241

Open yonjah opened 5 years ago

yonjah commented 5 years ago

I see some abandoned projects with only a few releases getting a maintainability score of 100

I think this is probably because evaluateReleasesFrequency and evaluateCommitsFrequency Only consider the time between commits / releases but do not actually consider the current date. So if last release was two years ago a project can still have 100% maintainability score. If on the other hand the same project would now fix some bug and add a new release they will actually loose their maintainability score.

If you consider the fact that forks and projects without any issue will get an issue distribution score of 0.7 where this score seem to average around 0.4 with some of the top maintained packages on npm a dead fork with a single release of any package can get a higher maintainability score than the original package

yonjah commented 5 years ago

Sorry just noticed #231 Though it seem almost any package with v >= 1 will hit the finalize check especially if it doesn't have anyone using it to open any issues.

Not sure but maybe a check if a package had a couple of dozen closed issues before assuming it's been finalized will improve results ?