I like the ingestion via bookmarklet method, as it's low user overhead; just does the job and gets out of your way.
I have 73k bookmarks on my instance, which other sass sites sometimes aren't optimized for, and can be sluggish while filtering searching etc.. (users normally have around 5k)
I can add detail for a docker build if desired, but the build from source is pretty straightforward.
I would love details on a docker build for my own homelab. I don't know Haskell, but perhaps it's time to tinker with something new.
SQLite's been excellent. Postgres wouldn't be hard to switch to in the future but I don't think I need it yet.
My initial need was met by delegating archiving to http://archive.li . Click the green checkbox next to the bookmark to access the archived page . While not a full mirror, I like having them as the permanent custodian.
I'll probably work on getting a release up that doesn't require a build from source such as docker or bits pretty soon.
In that context open source should be like a bazaar where we share ideas and compete a bit, and see what works, and learn from each other. Let a thousand flowers bloom; we aren't talking about seed rounds here.