Hello,
A bug was recently included in a public release as a result of contribution which was then reverted.
A simple auto-test would have prevented this bug from reaching the release.
Example test case: create test image containg all kind of filesystem objects: files, directories, symlinks (absolute & relative, both for existing and non-existing objects of all kinds), using all kinds of permission modes, etc. Then export the image to a new tarball and compare it with the original one.
This would make regressions of this kind much less likely.
Hello,
A bug was recently included in a public release as a result of contribution which was then reverted.
A simple auto-test would have prevented this bug from reaching the release.
Example test case: create test image containg all kind of filesystem objects: files, directories, symlinks (absolute & relative, both for existing and non-existing objects of all kinds), using all kinds of permission modes, etc. Then export the image to a new tarball and compare it with the original one.
This would make regressions of this kind much less likely.