This is a custom ejabberd module which allows users to acknowledge/retrieve their unread messages from direct chats and multi user conferences. You can implement things like the WhatsApp read message markers, notification streams with the ability to mark a single notification as read, etc., with this module. You can find further details of the concept to learn more about the client usage. This project comes with a self-contained test setup with all required parts of the stack.
Heads up! ejabberd-unread is the successor of the ejabberd-read-markers module, which differs in support for direct chats and per-message acknowledgements.
- ejabberd (=18.01)
- PostgreSQL (>=9.6)
- GNU Make (>=4.2.1)
- Docker (>=17.09.0-ce)
- Docker Compose (>=1.22.0)
See the detailed installation instructions to get the ejabberd module up and running. When you are using Debian/Ubuntu, you can use an automatic curl pipe script which simplifies the installation process for you.
We make use of the global database settings of ejabberd, but you can also specify a different database type by setting it explicitly.
modules:
mod_unread:
db_type: sql
Keep in mind that this implementation just features the sql
database type,
and only this.
The concept outlined the unread_messages
table definition which is required to
store the unread messages per user per conversation. The actual SQL
schema MUST be executed on
the Jabber service database (PostgreSQL).
The project bootstrapping is straightforward. We just assume you took already care of the requirements and you have your favorite terminal emulator pointed on the project directory. Follow the instructions below and then relaxen and watchen das blinkenlichten.
# Installs and starts the ejabberd server and it's database
$ make start
# (The jabber server should already running now on its Docker container)
# Open a new terminal on the project path,
# install the custom module and run the test suite
$ make reload test
When your host mDNS Stack is fine, you can also inspect the ejabberd admin
webconsole with
[email protected]
as username and defaultpw
as password. In the
case you want to shut this thing down use make stop
.
If you running Ubuntu/Debian, all required packages should be in place out of the box. On older versions (Ubuntu < 18.10, Debian < 10) the configuration is also fine out of the box. When you however find yourself unable to resolve the domains or if you are a lucky user of newer Ubuntu/Debian versions, read on.
Heads up: This is the Arch Linux way. (package and service names may
differ, config is the same) Install the nss-mdns
and avahi
packages, enable
and start the avahi-daemon.service
. Then, edit the file /etc/nsswitch.conf
and change the hosts line like this:
hosts: ... mdns4 [NOTFOUND=return] resolve [!UNAVAIL=return] dns ...
Afterwards create (or overwrite) the /etc/mdns.allow
file when not yet
present with the following content:
.local.
.local
This is the regular way for nss-mdns > 0.10 package versions (the
default now). If you use a system with 0.10 or lower take care of using
mdns4_minimal
instead of mdns4
on the /etc/nsswitch.conf
file and skip
the creation of the /etc/mdns.allow
file.
Further readings
- Archlinux howto: https://wiki.archlinux.org/index.php/avahi
- Ubuntu/Debian howto: https://wiki.ubuntuusers.de/Avahi/
- Further detail on nss-mdns: https://github.com/lathiat/nss-mdns
The test suite sets up a simple environment with 5 independent users. (admin, alice, amy, emma and bob). Multiple (9) scenario seeds gets created to test all features of the module. We create multiple MUCs, send messages to these MUCs and also send direct messages to configured conversations between two users. The MUCs are owned by the admin user, the message sender on all MUC messages is alice. The sender of direct message varies. The suite performs then all tasks defined by the concept on the service, which includes marking a message as read, fetching all unread message counts of all user conversations, etc. The test suite was written in JavaScript and is executed by Node.js inside a Docker container. We picked JavaScript here due to the easy and good featured stanza.io client library for XMPP. It got all the things which were needed to fulfil the job.