It's fun to test the boundaries of github's services, but if you're doing something useful I'd just hire a vps, they can be had from $5 a month. You could still upload the sqlite file to github via a check-in.
Presumably you can bypass the artifact retention limit by uploading them as release artifacts (which are retained forever) rather than job artifacts.
(Not that I’d advocate for this in general, since ultimately you’re duplicating a bunch of data and will eventually catch the eye of some GitHub compliance script.)
out of curiosity: is there a specific reason to use robinraju/release-downloader@v1 over actions/download-artifact@v4 here at your 'Download previous DB' step in build_db.yml?
It's used to download the previous day's database from the release for computing the diff. actions/download-artifact would only work for artifacts created during the current run, i.e., today's database.
interesting! perhaps cleaning up the older data might help abit here
> since ultimately you’re duplicating a bunch of data and will eventually catch
the eye of some GitHub compliance script
I suppose this could also be a concern with git scraping as we are bascially duplicating data through git commits (not trying to imply that one is better or worse). Having that said, I'm not sure if GitHub would be fine with any of these if more people were to do the same at a larger scale