What is CR-Avatar?

CR-AVATAR lets a registered person pick one emoji (from a fixed list of 184) to stand next to their name. It is text only: no pictures, no URLs, no uploads. On ConferenceRooms you see it in the Web Chat, and in IRC programs that support the protocol, such as Textce on Mac, iPhone and Windows. Everywhere else the emoji still shows as one line in /whois.

Four parts, four jobs

  person types  /avatar fox
        |
   [ IRC services ]   keep the emoji with the registered account            -> services authors
        |  server link: a services tag (SVSTAG)
   [ ircd ]           knows each user's avatar and tells clients about it   -> ircd authors
        |  005 token  +  capability  +  AVATAR messages  (and the /whois line, always)
   [ IRC client ]     shows the emoji in the user list, in chat, in /whois  -> client authors

• Services are the source of truth. Clients never talk to services about avatars.
• Only the ircd can put new lines on a client's connection. So detection, the capability and the AVATAR messages are ircd work, and need no change in services that already send the avatar tag.
• Clients that do not know CR-AVATAR notice nothing, except the /whois line.

What goes over the wire

Clients behind a bouncer

A bouncer only passes what it understands. ZNC passes the CR-AVATAR=1 token but does not offer the capability, so a client behind ZNC falls back to polite, quiet /whois lookups (about one every three seconds, silent, only when the network supports CR-AVATAR). That is slower but works.

Where next

ircd-hybrid module (download, install, tests), services authors, client authors, and the full specification.