SafetyError: The package for libboost located at /Users/runner/miniforge3/pkgs/libboost-1.82.0-h72cdd8a_4
appears to be corrupted. The path 'lib/libboost_atomic.dylib'
has an incorrect size.
reported size: 87312 bytes
actual size: 86178 bytes
SafetyError: The package for libboost located at /Users/runner/miniforge3/pkgs/libboost-1.82.0-h72cdd8a_4
appears to be corrupted. The path 'lib/libboost_chrono.dylib'
has an incorrect size.
reported size: 61328 bytes
actual size: 60194 bytes
SafetyError: The package for libboost located at /Users/runner/miniforge3/pkgs/libboost-1.82.0-h72cdd8a_4
appears to be corrupted. The path 'lib/libboost_container.dylib'
has an incorrect size.
reported size: 123552 bytes
actual size: 122421 bytes
and so on...
This seems to have been overlooked in #164, unfortunately. š
We did have some other SafetyErrors for the unification related to CMake metadata, which got fixed in 43373e65b5721f029fd59db95cf9ab7cd74dc9b8 and 3c87dcf6d04df7b904c46d5267459d5a78a4624d; this could be similar in principle.
What I could imagine happening: we're generating signature with cctools upon first installation (for libboost), which changes the artefacts. Once we copy them again for libboost-devel, the signature doesn't get generated, and so conda gets confused with the sizes?
Just noticed this in CI:
and so on...
This seems to have been overlooked in #164, unfortunately. š
We did have some other SafetyErrors for the unification related to CMake metadata, which got fixed in 43373e65b5721f029fd59db95cf9ab7cd74dc9b8 and 3c87dcf6d04df7b904c46d5267459d5a78a4624d; this could be similar in principle.
What I could imagine happening: we're generating signature with cctools upon first installation (for
libboost
), which changes the artefacts. Once we copy them again forlibboost-devel
, the signature doesn't get generated, and so conda gets confused with the sizes?