Philosophical question regarding file permissions #239
Description
We've been investigating some weird issues between our CI environment and certain development environments (ahem, CodeSpaces). We've been seeing that the CodeSpaces environment produces different hashes than everywhere else, especially for python_binary (pex) targets.
After some investigation, we determined that the problem is that the pex files produced in each environment have different permissions applied. This appears to be caused by an ACL configuration issue in CodeSpaces that cause them to ignore umask. If you remove the ACL and adjust all the permissions, everything works as expected.
The reason I opened this issue for feedback is simply to ask whether or not Please should be normalizing those permissions as it builds the archive, or if that's something we need to ensure elsewhere. Since git doesn't track file permissions except the owner execute bit, perhaps please could do the same? Or would it make sense to provide a flag that controls this behavior somehow?