Hey everyone,
I resumed work on my Elixir <=> SQLite library (which uses a Rust NIF underneath) and I am in a need of small and well-structured open datasets, preferably directly saved as an SQLite DB – but they could very well be everything else, as long as they can be easily and quickly converted to SQLite.
I downloaded the much-known Chinook DB but it’s a bit too big for my tastes. I plan to bundle a small DB with my library so end-to-end integration tests can be ran on it and that means that the DB must be extremely minimal (no more than 100KB).
Lately did some Rust code reorganization. Next up is finding a minimal locking implementation since sqlite3 is quite liberal with locks (but even it can break if several very particular operations are done in parallel – even if they are allowed by sqlite3).
Still, for an MVP I want something that’s with zero locking lag.
To be fair, the Elixir part seems quite easy – albeit laborous. I am prioritizing making the Rust code to be the best possible first.