khepri
khepri copied to clipboard
Unicode in paths
Currently, Unicode in paths is undefined behavior or defined as crashing behavior in some places :-).
It would be nice to improve this, at least to have a clear behavior even if it means some things are unsupported.
Unicode affects:
- the conversion back and forth between native and both string- and binary-based Unix-like paths
- name matching in the
#if_name_matches{}and#if_path_matches{}conditions.
We can't really settle on e.g. UTF-8 encoding for binaries because it will impact users who put non-text binaries in node names. I didn't make my mind on this issue yet and would welcome any ideas!