opam
opam copied to clipboard
Regression on `opam admin cache` incremental runs
The feature was added in #4939 and seems to no longer work, I get lots of mismatched checksums even without the option --check-all.
There was some refactoring in checksum handling and OpamRepo, maybe the check is now done more eagerly and isn't properly bypassed ?
Apart from speed with small cache updates, a use-case is to use a cache on git LFS where the files may be placeholders.
It actually seems to work on new successive runs; only broken on older caches that were based upon sha256/md5 checksums ?