Skip to content

netdoc: specify utf-8 behavior

Nick Mathewson requested to merge nickm/torspec:utf8-status into main

This spec change is a bit of a bodge, since Arti always enforces UTF-8, but C tor only enforces it in a couple of particular cases. Neither one exactly matches what proposal 285 had envisioned.

FWIW, at no point ever did C Tor or anybody else enforce the rule that Arguments had to consist of printable ASCII characters. Before we decided that our encoding would become UTF-8, the only rule that C tor followed was that Arguments couldn't contain NL or NUL. All other byte sequences were accepted.

See also arti#1739 for bring Arti's behavior into conformance.

See also #297 for the impact of changing which descriptors are acceptable.

Edited by Nick Mathewson

Merge request reports

Loading