Always wanted to have something like the crafting interpreters style book but implementing a database from scratch, and bonus points for a distributed system with a leader/replica model, leader election and things like partitioning, resizing partitions, handling node failure.

Something like building a toy dynamodb variant. Would pay good money for this.

Is one thing to read about and guesstimate implementation choices from white paper. Totally another if databases/distributed systems expert walks you through propagating write ahead logs to replicas. Hehe

And what it means to have eventual consistent writes vs strong writes in practice. A way to teach by doing.