* Setting video owner as room owner as side effects. Like leaking the instance moderator/admin list. And maybe others.
* So we rollback this feature, waiting for a better way to allow room admins/moderators to change room settings.
* As a consequence, users that are not Peertube moderator/administrator can't change the slow mode configuration without deleting/recreating the room. This will be fixed in a later plugin version.
* This was related to the fact that the bot was owner because of the "admins" Prosody config key, and not registered in the room.
* To fix it, i added the bot as owner when creating rooms.
* This means that the fix only work for newly created rooms.
* There might still be an issue in ConverseJS or Prosody, don't know exactly where.
* adding new avatars by David Revoy
* replacing old avatars to test
TODO: add a settings to choose between legacy and new avatars.
TODO: generate a bot avatar.
* For now, only one avatar possible,
* uses mod_random_vcard_peertubelivechat,
* mod_random_vcard_peertubelivechat has a new option with the avatar
list (instead of a hardcoded avatar number)
* Peertube lists available avatars files, and pass it to mod_random_vcard_peertubelivechat
* New settings: "Ban anonymous user's IP when user is banned from a chatroom":
* if enabled, every time a streamer bans an anonymous user, it will ban its IP on the chat server,
* banned IPs are logged on disk, so server's admin can use them to feed fail2ban (for example),
* option disabled by default, because could be used to create trapped-rooms on public servers
* Start and stop the bot WIP
* Prosody: removing the BOSH module from the global scope (must only be present on relevant virtualhosts)
* Some refactoring
Bypassing Nginx for API call originated from Prosody (if Peertube >=5.1).
Can also fix some Docker setup, which needed to set the prosody-peertube-uri settings.
Letsencrypt certificate import procedure was not working on server that had never installed Prosody.
Adding `prosody_user` in the Prosody configuration file to fix this.
Updating the procedure: the `chown` is no more needed in `/etc/letsencrypt/renewal-hooks/deploy/prosody.sh`.
Trying to connect to a remote instance using direct s2s won't work if local instance has not the feature enabled, and if the remote instance does not know the local one.
So using Websocket S2S in such case (that embed a discovery mecanism).
This commit is a dirty hack to fix issue with ARM64 Prosody AppImage.
The fixed appimagebuild tool is not released by the dev team.
So i patch it in my build script. This is not a good solution, but it
works, and let me go forward.
Note: websocket s2s is not working yet, still WIP.
New Features
* Chat Federation:
* You can now connect to a remote chat with your local account.
* This remote connection is done using a custom implementation of [XEP-0468: WebSocket S2S](https://xmpp.org/extensions/xep-0468.html), using some specific discovering method (so that it will work without any DNS configuration).
Minor changes and fixes
* Possibility to debug Prosody in development environments.
* Using process.spawn instead of process.exec to launch Prosody (safer, and more optimal).
* Prosody AppImage: fix path mapping: we only map necessary /etc/ subdir, so that the AppImage can access to /etc/resolv.conf, /etc/hosts, ...
* Prosody AppImage: hidden debug mode to disable lua-unbound, that seems broken in some docker dev environments.
* if both local and remote instance have external XMPP connections enabled, the user joins the remote room with his local account
* some code refactoring (builtin.ts)
Note: documentation and settings descriptions are to do.
Related to #112
* new option to allow external xmpp account s2s connections
* settings to setup the port and network interfaces
* documentation WIP
Related to issue #114
