Unexpected behavior on trailing slash

[TheDaemoness]

My intuition is that paths with trailing slashes should be treated the same as paths without them. I've written my EditorConfig core in Rust (which, at the time of writing, passes the full core test suite) to have this behavior.

This does not match EditorConfig Core C's behavior.

$ cat .editorconfig
root = true

[foo]
matched=true
$ editorconfig --version
EditorConfig C Core Version 0.12.5
$ editorconfig $PWD/foo
matched=true
$ editorconfig $PWD/foo/
$ ec4rs-parse $PWD/foo
matched=true
$ ec4rs-parse $PWD/foo/
matched=true

[cxw42]

That is an interesting one. I recall that the C core assumes its inputs are fully canonicalized, which I don't think $PWD/foo/ is. I can't find a clear statement on the subject, but I did find some data:

• POSIX 2004 A.4.11 "requires that a pathname with a trailing slash character be treated as if it had a trailing /. everywhere." Since canonicalization strips /. components, that would imply to me that trailing / was not canonical.
• POSIX 2017 4.13 says that, for symlink resolution, resolution is not complete if the pathname has a trailing slash.