Roadmap for 2018

Hey everyone!

As first half of the year nears the end, I would like to share updated roadmap for the rest of 2018:

Social Auth

Currently under development, "Social Auth" means integration of python-social-auth into Misago's code, thus making it possible for site administrators to enable users to join or log in using their account in social site supported by aforementioned library.

For power users python-social-auth exposes the framework for writing custom integration with oauth poviders that are not supported in the core.

UPDATE: Social auth has been released in Misago 0.18!

GDPR

As soon as social auth gets released, I'll be moving on to implement features making it easier to comply with GDPR regulation that comes into effect in EU on 25th of May. Misago already implements data anonymization on account deletion, but there are two other tasks imposed on site administrators that could be supported in core package: adminable list of explicit consents and personal data export.

UPDATE: GDPR has been released in Misago 0.19!

Docker for local dev and production

We want to make it easier to start and run Misago communities as well as to try it out locally to for those who want to see it's moderation and administration tools. Thanks to contributions from @einarf, it is now possible to get Misago running on your machine with zero file edits and no python knowledge by just 3 commands in terminal.

The next step is to provide Docker setup that can be ran in production environments, simplifying long and tedious process of setting up and configuring Misago, UWSGI, Nginx, Lets encrypt with caching and email, as well as abstract away the process of editing settings.py. Instead user will be provided with basic config file and guide on how to go about different setups, with actual configuration using sane defaults being generate from that. Power-users will still be able to get their hands dirty with configuration files.

UPDATE: As of 02.11.2018 Misago-Docker is available for production use.

Removing Python 2.7, Python 3.4 and 3.5 support

Increasing number of Python libraries are becoming Python 3.5-only, with special attention being given to Python 3.6 that brings async facilities and useable type annotations to the Python. I think its time to drop 2.7, 3.4 and 3.6 from Misago codebase, to open the way for those libraries, as well as to reduce current maintenance burden.

Single Sign On/Integration with existing sites

Top feature request of all time for Misago is easy integration with existing sites. While my original plan was to try make Misago easier to integrate with existing Django projects, after much exploration on the matter I have concluded that doing so would be a mistake, resulting solution nightmare to maintain and integrate with.

Instead Misago will implement features enabling it to be configured to use cookie with JWT set by your site as well as to redirect users to your login/registration forms, as well as expose you additional API endpoint to forcefully update user from your own application if you wish so.

GraphQL API, deprecation of JSON API

When Facebook was demoing the GraphQL on ReactConf 2016, I've considered it interesting but terribly immature piece of technology. The time sure flies, and GraphQL together with tool ecosystem has matured a lot in those 2.5 years.

Today I am convinced that GraphQL is the way to build API for modern web applications, and I want to replace current JSON API with new GraphQL API in one of future releases, making it a new requirement for new frontend.

New frontend

My grand vision for Misago is software package with two components: the server written in Django and exposing powerful API for building internet forum, and client written in TypeScript with React.js. If you want to customize looks, you can either fork default client, or write your own in whatever technologies you are comfortable with. The server will implement only the basic templates for clients that don't support JavaScript, such as some internet crawlers or web scrapers. It will also come with default client included so you don't have to install, build and deploy it separately to get started.

Due to large scope of this change, it is unlikely to land in 2018, but it will also not be a blocker from making smaller maintenance releases adding small tweaks, fixes or features.

I know that both NodeBB and Discourse are providing client app for mobile that you can take for free then release on app store for your users to use to access your community instead of responsive version. It's supercool, but also something that takes a lot of effort to get done, which makes it very unlikely to happen for Misago unless somebody else contributes it to the project.

As usual, things are released on the "when it's done™" basis, with no ETA being given until feature enters testing phase ;)

Social Auth is basically done and has been deployed on this site for testing. I just want to add and test Facebook Login before releasing.

Nothing set in stone yet, but recently I'm feeling that roadmap should be updated as Misago is increasingly needing the quality of life improvements for site administrators and developers. Especially the former would profit the great deal if we provided sort of docker-based setup together with useful tools and documentation for deployment and maintenance tasks:

• perform some required configuration on fresh server
• setup all services needed to run Misago (Python, UWSGI, PostgreSQL, Nginx, Let's encrypt, Crontab, Memcached, Mailing)
• update Misago to current version (update settings.py and urls.py, collect static, run migrations)
• backup and recover the Misago if things go wrong

We could then stop writing update instructions as they are now, instead providing the release notes in our docs for power-users running custom deployments, and telling everybody else "just git pull the latest misago-docker (or your custom fork), run ./misago-cli check and, if its all-green, run ./misago-cli update".

At first, I've wanted Misago to be super minimal and dumb with its dependencies, but this approach is limiting to my ambitions and features I would eventually like to provide in Misago. Celery is a bit of technology that is super python/Django centric, but would open the door to super cool features and experience improvements like watching categories generating user-posted link previews or making it faster to reply to posts. By making people use misago-docker for deployments, celery and task que setup could happen automatically, in the docker, and without needing for manual configuration by the admin.

I've updated the roadmap, reordering the items and adding new ones.

We'll soon start October, making it 3 months to end of the 2018. Its already obvious that not everything mentioned in OP will be shipped this year. Giving list a quick skim, I think its likely that by end of the year we'll have SSO support, move our codebase to Python 3.6/3.7 and have misago-docker to simplify setup and running Misago in the wild. GraphQL and new frontend are too big of a task to happen in 2018, and will have to wait to 2019. 😱

Hi @Invisible

Thank you for your suggestions. While I am planning to include JWT support in Misago, this will be in form of SSO cookie that can be set by one site in domain, and then read by Misago running in same domain to create/update user account. GraphQL API will likely continue to use Django Sessions ;)

I've also skimmed those libraries and it seems that they all are for Graphene. I also don't want to spoil the surprise, but Misago will not be using it for its GraphQL API ;)

I can't wait to start working on it, but need to land GraphQL first. And for GraphQL I need to update our server code a little.

I will be also deprioritizing the "SSO" cookie support, Misago 0.20 will instead include admin interface for easier customization.

Thanks for feedback @rokkirokki. Could you provide more details on what problems or headscratchers with customizing Misago look exactly are you having now?

I agree that this field is lacking currently, but different folk have different backgrounds and problems, so I'm trying to get a clearer picture here for how can I make this easier for you and your team.