I saw in the docs that it uses IndexDB. Didn’t read carefully to understand how full is the replica and if it’s possible to make 100% offline app and what would be the limits of storage in the browser.
Also it would be nice to clarify memory consumption.
You could theoretically make a 'fetch all' query, and replicate a completely offline experience. However, Instant is designed for hybrid use cases.
> what would be the limits of storage in the browser.
The limits are set by IndexedDB, which are a bit esoteric (it depends on how much space the user has available on their hard drive). It could reach into the GBs, but then the browser can sometimes choose to delete it. This comment goes more into it: https://news.ycombinator.com/item?id=28158407