PersistentSortedSet documentation

Deep-dives behind the features summarized in the project README.

Storage & durability

• Serialization — the optional canonical Fressian read/write handlers for nodes and roots (pss/leaf, pss/branch, pss/set); self-describing blobs; resolver registries for shared/wire serializers.
• Diff buffering — write-amplification reduction on content-addressed storage: bring a content-only commit down to ~1 object by buffering each unchanged-structure child's diff at the serialization boundary. Model, invariants, on-disk format, storage contract. Opt-in, off by default.

Content-addressing & sync

• Merkle Search Tree (content-defined boundaries) — experimental, prolly mode: split decisions derived from key hashes so that equal sets are byte-identical trees, independent of operation history. For CRDT state sync, cross-replica dedup, and O(1) idempotence. Opt-in, off by default.

Queries & statistics

• Statistical queries — subtree counts and the monoidal :measure mechanism behind count-slice, get-nth (rank/percentile access), measure, and measure-slice.

Internals

• B-tree operations — the split/merge/borrow mechanics, transients, and the node-level invariants shared by all boundary modes.

---

For the public API and worked examples, start with the README. The version history is in CHANGES.md.