Expectations vs RDBMS

  • Schema & metadata — “Tables” may be collections, keys, buckets, or indexes. Entity lists might be discovered lazily or approximated from samples.
  • SQL — There is no universal SQL layer; ad hoc query support is provider-specific (e.g. Flux for InfluxDB).
  • Migrations & DDL — RDBMS-style migration helpers in BeepDM often do not apply; prefer provider APIs and contract methods the core actually implements.
  • IDataSourceHelper — Unlike RDBMS, there is typically no shared dialect helper tree; logic lives in each core.

By operational model

ModelExamples (this repo)Notes
Document (server)MongoDB, RavenDB, CouchDBBSON/JSON documents; sync and revision semantics vary.
Document (embedded file)LiteDBAlso under local & in-memory; NOSQL category.
Key-value / cacheRedisImplements IInMemoryDB in addition to IDataSource; key patterns not tables.
Time seriesInfluxDBMeasurement / tag / field model; token and org/bucket connection fields.

Dedicated graph categories may appear in other phases if cores register separately; current NOSQL cores in this wave are document, KV, and TSDB.

Contrast: REST / SaaS connectors

SaaS REST drivers under Connectors/ use DatasourceCategory.Connector, WebAPIDataSource, and WebAPIConnectionProperties — not document/KV wire protocols. See REST / SaaS overview and flagship conn-* pages.

Documented providers

  • MongoDB — MongoDBDataSourceCore/
  • Redis — RedisDataSourceCore/
  • RavenDB — RavenDBDataSourceCore/
  • CouchDB — CouchDBDataSourceCore/
  • InfluxDB — InfluxDBDataSourceCore/

BeepDM skills

  • BeepDM/.cursor/idatasource/SKILL.md
  • BeepDM/.cursor/universal-general-helper/SKILL.md (when a core delegates to a general helper)

Next implementation family: Cloud & analytics (DatasourceCategory.CLOUD).