The migration of from to is almost complete.

CI/CD pipelines aren't operative yet, but everything else should be working. And we've also got a "sign in with " button that uses my instance - how cool is that?

Gitlab used to take about 5 GB of RAM and a lot of CPU to run just the basics. Gitea takes barely 100 MB of RAM and it's so light on the CPU that it could even run on a Raspberry Pi. Thanks to this migration alone I've managed to downgrade my 16 GB VPS to 8 GB and save about $60/month, while having basically the same features (except for integrated CI/CD) and a much faster response time. I should have probably done this earlier.

There's only one heads up: in the process of migrating the users I couldn't (obviously) decrypt the hashed passwords, so users are required to use the "forgotten password" link to reset it, or use one of the available SSO methods. I'll probably send a communication soon to all registered users.

After my article on how to create / -> cross-posting bots, I did an experiment with @crossbot and let it run with ~10 different sources for a couple of weeks.

The idea was definitely successful: I brought with me to the Fediverse all the sources that I wanted to follow, without forcing them to move, and I actually didn't feel the urge to open Twitter/Facebook for "fear of missing out".

But I've realized that one single bot to manage multiple sources isn't ideal. People who may want to follow only some of them are forced to get on their timelines also content that they didn't ask for. Some people did indeed follow crossbot, but many also unfollowed it - probably because it posted too much, too often, and since all the content was coming from the same account it was hard to tell which was the source without actually reading the toot.

So I've decided to split it into multiple bots, one for each of the sources that I'm cross-posting. Feel free to follow any of these bots if you are interested in the content! But please also avoid commenting on their activities (there's no human behind the profile that can react). Instead, favourite/boost/re-share the link if you want to bring the discussion to the "human" sphere.

List of available bots:

- The Economist: @economist_bot
- Quanta Magazine: @quanta_bot
- Nautilus Magazine: @nautilus_bot
- Nature: @nature_bot
- Scientific American: @sciam_bot
- @physorg_bot
- The Gradient: @gradient_bot
- The Hacker News: @hackernews_bot
- Hackernoon: @hackernoon_bot
- IEEE: @ieee_bot
- IoT for All: @iot4all_bot
- Better Programming: @better_programming_bot

Also, feel free to comment on this post if you have any requests for interesting sources that are only available on Twitter/RSS and you'd like to bring here - I may definitely consider making a bot for them.

It took me a while, but I've finally managed to get my working on all of the servers ๐ŸŽ‰

You can now register and login with a single account on:

- The instance (
- The instance (
- The instance (

And I've also got it configured on my instance (it was the most painful one to configure), though through a different realm so I won't get random people poking in my personal cloud :)

Things I've learned in the process:

- SSO on your network is amazing! You can also configure WebAuthn over e.g. Yubikey, and you won't have to keep track of tons of different credentials across several services. No Google/Facebook/Microsoft or any third-party SSO solutions required.

- is the most popular and complete solution, but it's unjustifiably heavy. It's maintained by Red Hat, it runs on top of an obese web server like , it takes at least 1 GB of RAM to run, it comes with tons of unintuitive configurations, and it screams "heavyweight enterprise sh*t" from everywhere. Wish there was a more lightweight solution with a less steep learning curve.

As a user, I can follow a lot of cool people, but I can't access content that is exclusively published over Twitter.

Until recently I still opened to check for updates by profiles such as MIT Technology Review, The Gradient, The Economist, Quanta Magazine or Phys, since none of those accounts cross-posts to the Fediverse.

That's no longer the case. I decided that instead of complaining about the mountain not moving to me, I should probably take the initiative and drag it myself.

So I have created a based on (and a sprinkle of ) that subscribes to a curated list that contains my feeds and with my favourite Twitter accounts (using nitter to bridge Twitter timelines to RSS), and forwards updates to my instance:

If you're into science and tech content, feel free to follow it!

And I've written a blog article that explains how to build a bot like this, together with some random thoughts on the Fediverse.


A platform about automation, open-source, software development, data science, science and tech.