Skip to content

Release Candidate

This page captures the current v0.1.0 release-candidate materials for the first public CypherGlot release.

Target version

  • release version: v0.1.0
  • package version source: Hatch VCS tags
  • source-file bump needed: none

CypherGlot already derives package versions from Git tags, and the current Hatch fallback version is already 0.1.0. For the first public release, the version step is therefore operational rather than code-facing: the release tag should be v0.1.0.

Release-candidate summary

CypherGlot is ready to present as a compiler-focused package, not as a database engine and not as a full-Cypher compatibility claim.

The release candidate is centered on:

  • repo-owned Cypher parsing
  • explicit admitted-subset validation
  • repo-owned normalization
  • SQLGlot-backed single-statement and multi-step compilation helpers
  • a practical mainstream single-hop, read-heavy onboarding subset
  • refreshed documentation and README positioning
  • a checked-in benchmark harness plus current summary artifacts

Changelog summary

The v0.1.0 changelog lives in CHANGELOG.md and should be used as the source summary for the GitHub release body and any PyPI release notes.

Tag plan

Recommended first-release flow:

  1. Confirm the release candidate is still green locally.
  2. Confirm CI is green on main, especially test.yml and build-docs.yml.
  3. Re-run the generated frontend check if parser artifacts changed: scripts/dev/regenerate_cypher_frontend_docker.sh --check.
  4. Ensure the checked-in benchmark artifacts still reflect the intended release state.
  5. Create the annotated Git tag v0.1.0 from the release commit.
  6. Let publish-pypi.yml publish the package from that tag.
  7. Let deploy-docs.yml publish the versioned docs and set the latest alias if appropriate.
  8. Create the matching GitHub release using the draft notes below.

Release notes draft

v0.1.0 is the first public release candidate for CypherGlot, a reusable, SQLGlot-backed Cypher frontend compiler for the HumemAI stack.

This release intentionally targets a practical mainstream single-hop, read-heavy subset for Neo4j-style onboarding rather than full Cypher parity. It includes repo-owned parsing, admitted-subset validation, normalization, and public compilation helpers for single-statement reads and narrow multi-step writes. The release also ships refreshed docs, a clearer public contract, and a checked-in benchmark harness with baseline corpora and current repeated-run summaries for future regression tracking.

Notably out of scope for v0.1.0 are broader MERGE semantics beyond the narrow admitted idempotent subset, broader variable-length path semantics beyond the bounded read-side subset, broader write-side traversal semantics beyond the narrow traversal-backed alias-reuse subset and one-fresh-endpoint MATCH ... CREATE / MATCH ... MERGE subsets, and backend-native compilation of vector-aware CALL db.index.vector.queryNodes(...) flows.

Pre-publish checks

  • uv run pytest
  • scripts/dev/run_postgresql_runtime_docker.sh
  • uv run mkdocs build --strict
  • python -m scripts.benchmarks.compiler.benchmark
  • scripts/dev/regenerate_cypher_frontend_docker.sh --check