elliebike

Yup you totally can! I've done this before + it works well. You may have to customise the DHT a bunch though

It's no different to most other social networks that commit mass abuse of user data and track everyone, everywhere.

Being a US-based company doesn't make this any better, just look at Facebook.

The only people who worry are people on HN. Most other people aren't aware of what's going on, or just don't care.

...really?

thank you! hope you're having a good day :)

I always feel weird about comments like this. Instead of discussing the merits of the actual project shared, we're going on a rant about whether or not ORMs are a good idea?

It's like whenever someone shares something they made with Electron, and the comments are just Electron hate.

I guess I got lucky then! I've had one of these for quite a long time, maybe since 2013/14. A good number of those years it's been running 24/7, and I actually only recently replaced it

Thank you! Yup, that's basically why I made it!

> More details on how the encryption for server sync works would be helpful for assessing security.

Good point - I'll add something more detailed to the README shortly

For now though, it's fairly simple. Almost all of the information is first serialized with MessagePack, and then encrypted using libsodium's secretbox. This blob is what ends up being POSTed to the server

The only other information available to the server operator is

  - the username/email address/hashed password of the user
  - the UUID of this history item
  - the history timestamp (needed for sync at the moment)
  - the hash of the machine's hostname (also useful for sync)

Hopefully this works well!

Honestly the multi-device support is a big barrier to my adoption of Signal. Currently I use Telegram across several devices, and the messages just "sync" near instantly. With Signal you see them appear one-by-one, which is painful in a large + busy group chat. Obviously Signal has technical issues to overcome that Telegram does not, but the UX issue is still there

Hey HN! I wanted to learn about training + deploying ML models, so I built this!

It uses YOLOv4 to find license plates, and then opencv to blur them :)

Happy to answer questions!