The Merkle Search Tree paper is pretty cool
@makeworld I'm abour to publish a thing that goes over the tradeoffs between it and prolly trees actually!
@mauve haha that's perfect, I was just wondering that
Follow
@makeworld tldr is MSTs guarantee at most one chunk boundry update on insert, prolly trees have a (rare) chance to cause "cascade" of chunk boundry updates. Prolly trees are easier to construct when you have sequential keys and do less lookups for sequential reads. MSTs also have more even distribution of items accross layers.
@makeworld the origibal noms doc about it is great: https://github.com/attic-labs/noms/blob/master/doc/intro.md
@mauve thanks! If you have any "intro to prolly tree" links you like I'd take those too :)