Git: list case-sensitive paths that have collided during clone

Question

When cloning a git repository that contains case-sensitive file paths (eg /README.md and /readme.md) on a case-insensitive file system (like NTFS or APFS), git will only check out one of the colliding files.

In macOS, how can I list all the files that collided because of case insensitivity?

Answer 1

There is no built in thing to find this. phd's comment will get you close, possibly close enough, but may over-fit a bit (although one might still like to know about these things).

For instance, suppose some commit has files:

path/TO/file1.ext
path/to/file2.ext

On your file system, only either path/TO or path/to may exist. Once one of those exists, these two files will be dropped into the same path/$to folder, where $to is either lowercase or uppercase. They will still be separate files, but will be called out by case-folding and sort-and-unique-dash-c-ing.

On macOS, we can also have collisions in paths due to Unicode normalization. Linux considers a file named 's' 'c' 'h' 'combining-umlaut' 'o' 'n' to be one file name, and a file named 's' 'c' 'h' 'o-with-umlaut' 'n' to be a second, different file name. The macOS default file systems will turn both names into a common form and claim that this is just one name. (I have no idea what Windows does with this.) A proper tool will should take this into account as well.

Note that Git will store each file separately in the index, and can update each separate index entry from a file-system-stored-file independent of the stored-file's path name. So we could have Git build a mapping from internal name to external name and make it handle these cases all automatically. But that's a pretty big task.