Move off MongoDB

[zkldi]

It's becoming more and more clear that the MongoDB setup we have is not good. It was an initial design choice made in Tachi 1 allll the way back in pre-covid times, but it was the wrong choice.

It was always the wrong choice, and although its served me well in quick prototyping... I now just want the site to stand on its own two legs without constant maintenance.

The truth of the matter is that the way we've engineered our database is beyond silly. I am never ever taking risks on my database decisions ever again. We'll be moving over to Postgres and all design decisions that come with that will be changed. This is something akin to a third rewrite of Tachi, as our use of MongoDB finds itself pretty bloody ingrained in the app.

I don't exactly want to spend another 3 weeks of my life rewriting Tachi (I'm working on other, more important things), but there are increasing failure rates in the app, and things are getting progressively less and less tenable. My hand is slowly being forced.

This won't happen for a while, but it is the one issue from which all things descend. This does need to be done, but it's not likely to happen this year.

I won't be accepting any PRs for this, as it is a full scale Tachi rewrite and most likely something only I'm equipped to do reasonably.

🫡

[zkldi]

Increasingly urgent; doing practically anything in this codebase is a nightmare because of how obscenely complicated the MongoDB "caching state" thing is. I've learned a lot.