You can set it up as a “public” instance with closed registrations, and basically just use it as a self-host but have a lot more visibility and control. Or, you could probably play around with the nginx config and make it so that only the federation endpoints are accessible to the world and the actual web app is limited to just you. It will need to be “public” in some sense in order to be reachable to receive the content from other instances though.
I personally don’t use or like docker, partly just from inexperience with it and partly because I like to have more hands-on control over the deployment than I’m able to get with docker.
Yeah, no admin is going to give you access to their database. Even if it is supposedly read-only or something, you would be able to read private messages and other things you really shouldn’t be able to read. There is also a theory that things like who voted for what are “supposed to” be private even though they are not. I don’t subscribe to that theory but that’s the prevailing view among Lemmy people I think. You would have to set up your own full instance which requires a fair investment of time and knowledge at this stage however you are doing it.
This doesn’t work though. It has to be sitting, subscribed to get API calls when people do votes, otherwise it won’t be able to find out any of the voting information. You can cause Lemmy to dump the comments or posts (up to a point) from a selected actor, but you can’t do that for votes. You’re either subscribed when the votes happen and available to get a call, or else it’s gone for good.
I know a little bit about it. I actually made myself use podman for a deployment not that long ago and I just didn’t like it. I think mostly the issue is just that I have not been in a position where I really had to do anything with it, but if I do wind up in one I may take you up on it.
Generally I like to muck around with the code for pretty much any service I am using / hosting. Just little tweaks to make things more amenable to how I like them. Telling docker/podman to do a source checkout and then recompile when I change something in the source in the container was beyond me and it didn’t really seem like it was set up with that kind of thing in mind, so I more or less abandoned it and went back to doing “from scratch” installations for any stuff I’m mucking around with.
Probably if I didn’t do that, I would prefer the simplicity / reproducibility and such that Docker gives, but that’s usually my priority over that.
Yeah, agreed. Like I say, I think it is fine as long as it’s clearly communicated to users that the expectation is that. I think without that clear communication, presenting a false sense of privacy, it’s wrong.