Hacker News new | ask | show | jobs
by interstice 291 days ago
This got me thinking a bit more. Rest / GraphQL / Groq handled with adapters, flatten anything nested that references an ID to the row level. Opinionated queries (queries only fetch a superset/subset of the same structure). Fetched data 'fans out' the new content into the rows based on ID to fill out/update structure. Lives in a service worker or side by side with frontend. Drops oldest/least fetched data when limits are reached. Would something like that work?

Alternatively just ship an entire shallow copy of least changed / most used data as sqlite db's to the edge, push updates to those, and fetch from source anything that isn't in the DB. Might be simpler.