b59c8f1f2c
- remove unused pg_cache build steps - add pgmq (Postgres message queue) build and installation - ensure pg_jieba build installs correctly via cmake - update runtime description to reflect pgmq inclusion - standardize build-stage layout and file copy structure
33 lines
1.1 KiB
Plaintext
33 lines
1.1 KiB
Plaintext
# TL;DR – PostgreSQL Multi-Model Runtime
|
||
|
||
A lightweight PostgreSQL build providing **Search + Vector + KV + MQ + JSONB**
|
||
as a unified data engine for CloudNeutral-Suite applications.
|
||
|
||
## Includes
|
||
- **pg_jieba** – Chinese full-text tokenizer
|
||
- **pg_trgm** – fuzzy search and typo tolerance
|
||
- **pgvector** – embeddings and semantic search
|
||
- **pgmq** – lightweight message queue (Kafka-lite)
|
||
- **JSONB + GIN** – document store and structured filtering
|
||
- **hstore + UNLOGGED tables** – high-speed key/value cache
|
||
|
||
## Use Cases
|
||
- Documentation, product, and FAQ search
|
||
- RAG and embedding-based retrieval
|
||
- Application-level KV/session/cache
|
||
- Lightweight event queues and workflows
|
||
- JSONB content and metadata storage
|
||
- Hybrid keyword + semantic search
|
||
|
||
## Not Included
|
||
Platform-level or DBA-oriented extensions are intentionally excluded:
|
||
- timescaledb
|
||
- pg_partman
|
||
- pg_cron
|
||
- pg_net
|
||
|
||
## Why
|
||
Keeps the runtime focused, predictable, and portable —
|
||
a single ACID engine replacing MongoDB + Redis + Kafka + Elasticsearch + Pinecone
|
||
for application-scale workloads.
|