Problems with non-ANSI characters in channel names

emacs-circe / circe

Circe, a Client for IRC in Emacs

GNU General Public License v3.0

390 stars 51 forks source link

Open MilkyPL opened 3 years ago

MilkyPL commented 3 years ago

After putting :channels (:after-auth "#cafè") in my config, circe would connect to #cafĂ¨ instead. Using the /join command manually works fine.

wasamasa commented 3 years ago

Seems to be an encoding issue. A similar problem has been found with passwords containing accented characters.

So what most likely happens:

Channel name is interpreted as UTF-8 encoded string (five codepoints, with è being one)
Connection to network is established, with the server supporting the latin1 encoding
UTF-8 encoded string is interpreted as latin1 (six characters, with è being turned into Ă¨)

What would need to happen instead is some sort of recoding.