They don't mention how the Output Gates interact with things other than the HTTP response. So it seems entirely possible that you could write some data, notify some other system with a POST message, and then try to respond OK to your caller... only to lose the write. Caller doesn't get his OK, but the other system DOES receive a POSTed message. Could be a different channel, e.g. websocket. Could be a write to D1, or KV. Many ways this leads to inconsistent state.

This looks absolutely amazing and absurd at the same time.

All IO is synchronous? Writes don't actually write when they say they do? Running 201 queries is as "fine" as running a single one with join?

It seems as they got every bad ideia in the book and somehow made it into something that works (that is, if it actually does).

The part where they query individual rows through SQL queries isn't "cringe" because it is inefficient. It is "cringe", because you have degraded SQLite to nothing but an overly verbose key value store.

Shared SQL databases tend to have very poor performance when you join across shards. In this case the downside of distributed joins is avoided by having very many small purpose built databases.

Honestly really impressive. SQLite is eating the world.