Instances can go down, disappear or be unresponsive due to many reasons.
Go to your account settings now and export your account settings!
This file contains your subscriptions, follows, profile settings etc. It’s very easy to start over on a new instance when you have your export file.
Back it up, export it, save it, repeat occasionally.
This simply needs to automatically happen every week or so.
How though? A website cannot just access your hard drive and write stuff, as far as I know.
Actually it can. Cookies, local storage … but using that for backups would certainly be an idea I haven’t heard before.
Plus it’s going to be a lot lire work to extract that when the website that wrote to it is no longer accessible. Easier to just click the backup button and download a file.
Mine is a 200kB text file (I imagine much smaller if compressed into e.g. a zip file), so it could just be attached in an e-mail 🤔
How about a desktop client?
Wait, doesn’t every website do this? Or are there websites that only live in RAM?
No websites should be able to access anything on your hard drive arbitrarily.
Websites can ask the browser to store something in cookies, or local db storage; it can even say “here’s a file for download”. But it cannot just decide to place a download somewhere in your filesystem.
The browser itself is accessing your filesystem to read and write cookies, cache, etc. and the website has (rightfully so) zero control over that, other than asking permission and offering up a link to something the browser may to may not decide what to do with.
Didn’t claim arbitrary access, just writing to hard drive. But I see, so the browser is the mediating layer
There’s a file api you can use to interact with the file system directly. Practically, I’m not sure how much it would help, since you could only access it with permission while the website is open, and there’s no way to sync between devices, but it is hypothetically possible to write a feature that backs up your settings periodically.
Automate a script to do it. It’s easier to make simple automations than most people would think
Both for the user and for the server/instance, it wouldn’t be wise.
From the perspective of instances, imagine having large instances (such as yours, lemmy.world, with almost two hundred thousand accounts as per FediDB current statistics) implementing a cron to compile and store a potentially large JSON/ZIP file for every account (including potentially inactive accounts), and having the storage requirements suddenly doubling (as the media files will become repeated twice in the server storage), which will make the storage quota/bill go through the roof for the instance owner(s) and/or, at best, having the Fediverse platform momentarily competing for storage resources with the backup cron. Notice I’m not just talking about the textual contents, but also about media (photos and videos) which should be included in the backup (otherwise the backup would be partial).
From the perspective of users, especially those who are prolific participants with thousands of posts, imagine having the instance pushing a large ZIP file into your browser’s (or phone’s, especially if you’re using a third-party app to access the Fediverse) storage every week or so, potentially in an non-consented manner, maybe pushing the backup media as new files so your gallery app (when in mobile environments) will get suddenly cluttered by potentially repeated images and videos.
Nonetheless, for most Fediverse platforms, the exporting feature is quite “automatic” already, as the backup file is often built in less than 10 seconds upon requesting it, but it only does so when the user requests so; given the unlikelihood that all users will request their backups at the same time, the backup feature (generally) doesn’t overwhelm the server, but it would if this automatic backup feature were a thing.