its an optimized, memory aligned binary format, with full local query capability, virtual memory mapped so you can page in gigabytes to terabytes of notes, with no synchronization or serialization required in/out, with an early exit json parser on duplicate notes for max performance. but yeah… “simple cache” 

 what did you use for inspiration? LevelDB? 

 strfry/lmdb 

 nostrdb is the same design as strfry pretty much and uses lmdb under the hood, except its an embeddable library like sqlite 

 Nice!!
Also sounds very familiar https://opfn.co

PLAN, in particular: https://docs.opfn.co/explanation/plan 

 ouch! 🤣