Skip to content

MemGQL: connectors + schema-graph model docs#1685

Open
DavIvek wants to merge 7 commits into
mainfrom
memgql-global-schema-docs
Open

MemGQL: connectors + schema-graph model docs#1685
DavIvek wants to merge 7 commits into
mainfrom
memgql-global-schema-docs

Conversation

@DavIvek

@DavIvek DavIvek commented Jul 10, 2026

Copy link
Copy Markdown
Contributor

Updates the MemGQL docs for the connectors + schema-graph model.

  • New Schema File page: boot MemGQL from one JSON file that declares connectors (connections) and the named graphs mapped over them; runtime DDL persists back to it.
  • Docs now lead with a schema-and-queries flow — declare your backends, boot, and query straight away. Routing picks the graph from a query's labels, so no USE clause is needed.
  • Documents the query-driven Memgraph cache (ALTER GRAPH … SET CACHE).
  • Quick Start, Reference, and the per-connector pages updated to the schema-file / Docker flow; single-connector env-var setup de-emphasized.

This PR can be merged once new version of https://github.com/memgraph/zero gets released.

DavIvek added 2 commits July 10, 2026 12:14
Update the MemGQL docs for the schema-graph model that replaces the old
connector-bound flow.

- New Schema File page: the --schema JSON (connectors + graphs, with the
  vertices/edges/attributes/metaFields mapping format), minimal + full examples.
- Reference: rewrite Graph Management to the real DDL surface (ADD CONNECTOR,
  CREATE GRAPH FROM, ALTER GRAPH SET READ/CACHE, SHOW ...); replace the old
  nodes/id_column mapping schema with a pointer to the Schema File page.
- Multiple Graphs, Quick Start, Docker Compose, and the public/private use
  case: registration, lifecycle, caching, and examples moved to the new flow.
- Oracle + SQL Server connector pages: ADD MAPPING / old mapping -> new flow.
- Changelog: Unreleased breaking-change + new-feature entries.

Removes all references to the retired statements (ADD GRAPH ON CONNECTOR,
ADD MAPPING, CONNECT ... AS, USE CONNECTION, SET DEFAULT CONNECTION, UNADD
GRAPH, ALTER GRAPH SET CONNECTOR/GRAPH/MAPPING) and the legacy nodes/edges
mapping format. Verified against the feat/global-schema binary (DDL, routing,
error cases, --schema boot).
Document the schema-file boot (docker run ... --schema=/data/schema.json) as the
simplest way to start a federated instance — in the Schema File "Booting"
section and as the "easiest" path atop Quick Start's Multi-Connection Mode.
Verified the binary honors --schema and BOLT_LISTEN_ADDR=0.0.0.0 (host-reachable).
@vercel

vercel Bot commented Jul 10, 2026

Copy link
Copy Markdown

The latest updates on your projects. Learn more about Vercel for GitHub.

Project Deployment Actions Updated (UTC)
documentation Ready Ready Preview, Comment Jul 11, 2026 2:00pm

Request Review

Format every JSON block (fenced code blocks + the mapping-file heredocs) with
standard 2-space indentation so the schema and mapping examples are easier to
read. Reformatting only — the JSON content is unchanged.
- Prefer USE-free queries throughout: drop unnecessary `USE` clauses and the
  USE-free explanatory comments. `USE` is kept only where it belongs — the
  query-model page (cross-backend joins, composites, disambiguation), the
  no-label cleanup query, and the reference. Verified USE-free single-graph and
  cross-graph UNION routing against a running instance (identical results to the
  explicit-USE baseline).
- Quick Start's Multi-Connection Mode is now schema-first: show a schema file,
  boot from it, query, then add more connectors/graphs at runtime.
- public/private demo: give the `friend_of` table an `id` column so the
  FRIEND_OF edge mapping (metaFields.id) and the insert/traverse steps work.
- Quick Start: the schema-file setup gets its own section, "Connecting your
  data with a schema file" (it was left under "Mapping File" when the
  multi/single-connector mode framing was dropped); the statements reference is
  now a top-level section. Mapping File intro no longer says "translates into
  SQL joins".
- Schema File: "How it works" trimmed to two conceptual bullets — dropped the
  hash-join / GQL->SQL / per-query-executor internals.
- Multiple Graphs: "cross-backend hash join" -> "cross-backend join".
- Reference: MAPPING_FILE marked required in the single-connector config tables.
…-dashes

- Rename the "USE-free routing" concept to plain "routing": section heading
  `## USE-free routing` -> `## Routing`, all `#use-free-routing` links ->
  `#routing`, and the term in prose (incl. historical changelog entries;
  terminology only, shipped facts unchanged).
- Booting from a schema file is now Docker-only; drop the bare
  `bolt_server --schema=...` invocation.
- Reduce em-dashes in authored prose across the schema-file, quick-start,
  multiple-graphs, reference, connector, and use-case pages. Data markers in
  SHOW SCHEMA/SHOW GRAPHS output and historical release notes are left as-is.
@DavIvek DavIvek marked this pull request as ready for review July 11, 2026 13:43
… link

The "Connecting your data with a schema file" quick-start section mapped a
social/User/FOLLOWS graph that didn't match the tutorial's seeded data, so the
sample query returned nothing. Map the seeded Developer/Language/WRITES labels
instead so the query returns rows, note it reuses the running Memgraph, and
reframe the runtime step as an illustrative example. Also repoint the
connect.mdx link from the removed quick-start#multi-connection-mode anchor.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant