Expand description
Storage primitives shared by native and browser runtimes.
The native pager uses these exact page, freelist, cache and byte-I/O implementations, including pager orchestration and allocation fences. WAL implementation and encryption remain host/runtime policy.
Modules§
- btree
- B+ Tree Implementation for Page-Based Storage
- context_
index - Context Index — dedicated cross-structure inverted index for context search.
- crc32
- CRC32 checksum implementation (IEEE 802.3 polynomial), SIMD-accelerated.
- entity
- Unified Entity Model
- entity_
cache - Sharded, bounded LRU cache for entity lookups by raw id.
- entity_
codec - Canonical logical entity adapter for native store versions V1 through V12.
Physical record framing and value bytes remain owned by
reddb-file. - entity_
id - Stable entity identity shared by storage and transaction tracking.
- error
- Canonical storage errors shared by native and browser hosts.
- freelist
- Free Page List Management
- manager
- Segment Manager
- metadata
- Type-Aware Metadata Storage
- metadata_
codec - Canonical native metadata bytes and entity-record adapter, shared with WASM.
- overflow
- Overflow chain storage
- page
- Page structure for RedDB storage engine
- page_
cache - SIEVE Page Cache Implementation
- page_
store - Page operations used by the shared B-tree and overflow algorithms.
- pager
- Shared pager orchestration over an exclusively owned VFS file.
- pager_
io - Byte I/O shared by the native pager and exclusively owned browser handles.
- segment
- Unified Segment System
- snapshot
- Streaming recovery of the existing native store dump.
- tokenization
- Shared tokenization utilities for indexing and search.
- visibility
- Visibility of persisted entities under an explicit statement snapshot. Host authorization, tenant and provider policy remain outside this module.
- visibility_
map - Visibility map — Fase 5 P2 building block.