ggen-engine 26.10.11

SPARQL-in-Tera code generation engine (vendored, renamed from praxis/crates/ggen)
docs.rs failed to build ggen-engine-26.10.11
Please check the build logs for more information.
See Builds for ideas on how to fix a failed build, or Metadata for how to configure docs.rs builds.
If you believe this is docs.rs' fault, open an issue.

ggen-engine

The live code-generation engine behind ggen sync: RDF/SPARQL ontologies in, Tera-rendered code out, with BLAKE3 receipt chains over every sync run.

Five-stage pipeline

ggen sync runs a sequential five-stage pipeline (crates/ggen-engine/src/sync.rs), each stage carrying its own OpenTelemetry span (pipeline.load, pipeline.extract, pipeline.validate, pipeline.generate, pipeline.emit):

  1. Load — load the manifest (ggen.toml), ontology, and referenced packs.
  2. Extract — run SPARQL SELECT/CONSTRUCT queries over the RDF graph.
  3. Validate — law/admission checks over extracted facts.
  4. Generate — render Tera templates from query results.
  5. Emit — write generated files; mint the sync receipt (.ggen-v2/receipt.json).

Graph backend

The default graph backend is the graphlaw crate (registry, pinned in the workspace), replacing the retired vendored praxis-core/ praxis-graphlaw members (SJIRA-15, 2026-10-10).

Admission gate

ggen-abb-sbb is wired in as the pipeline's Stage 2c admission kernel: IO-free ABB/SBB admit()/plan()/depgraph with transitive closure, cycle, and unbound-port refusals.

Related crates

  • ggen — CLI that routes sync/doctor/graph/ receipt here.
  • ggen-config — the ggen.toml manifest schema.
  • ggen-marketplace — pack registry/lockfile.

License

MIT. See LICENSE-MIT at the repository root.