[vidarh]

> More importantly, on CI systems it’s relatively common to check out code using git, and git doesn’t care about mtime

git doesn't care about mtime, but git maintains trees whose hash changes if any constituent part of the tree changes. It'd seem tempting to check for a .git and if present use the git tree to determine whether to invalidate the cache.

[byroot]

Aside from the oddness of making this cache git aware, with the new implementation I suspect querying git to revalidate the cache would take longer than just rebuilding it.

[vidarh]

Looking up the hash of a tree in git is few enough operations that I would be very surprised if that is true for all but the smallest caches. If you were to shell out to the git binary, maybe.