Stabilization before 1.0
The roadmap requires four actual weeks in the 0.9 series without a required breaking change, and zero unresolved critical bugs, before publishing 1.0. Documentation, passing tests or successful package publication cannot replace elapsed time.
The stable candidate comprises gritz-core, gritz-native, gritz, gritz-otel and gritz-rails 0.9.1. Async remains experimental, with a compatible 0.1-series release and its failed restart target recorded separately. The public API, support policy and security review define the candidate's scope.
All five 0.9.0 packages are published, with successful main and Trusted Publishing release workflows. Verification completed at 2026-10-03 09:11:28 UTC (18:11:28 JST); this is the observation start. The earliest possible 1.0 decision is 2026-10-31 09:11:28 UTC (18:11:28 JST). The four weeks have not elapsed, so 1.0 has not been tagged or published. A breaking correction resets the clock to the revised candidate's publication. Compatible fixes remain eligible only after their required checks pass.
| Gem | Version | Main CI | Release |
|---|---|---|---|
| gritz-core | 0.9.1 | Passed | Published |
| gritz-native | 0.9.1 | Passed | Published |
| gritz | 0.9.1 | Passed | Published |
| gritz-otel | 0.9.1 | Passed | Published |
| gritz-rails | 0.9.1 | Passed | Published |
| gritz-async (experimental) | 0.1.5 | Passed | Published |
All six versions and runtime dependency requirements were checked through RubyGems' individual version API; consumer gems require Core 0.9.1, and the meta gem also requires Native 0.9.1. The main suites contain 463 tests. Native and meta CI cover each of Ruby 3.3/3.4/4.0 with both minimum grpc 1.83.0 and the latest allowed version; Rails also covers 8.0/8.1. All six release workflows passed tests, lint and strict builds against published dependencies. The meta release checks the hello sample's four RPC forms; the Rails release checks real database RPCs on all four workers and process cleanup.
The documentation deployment passed, and the public site returned HTTP 200. Its public API inventory matched the reviewed candidate snapshot. The site build checks guide/reference pages, navigation and local file links.
The compatible correction releases on 2026-10-04 fix Native cooperative cancellation, readiness-timeout handling, missing operational commands and callback documentation. The adversarial validation records regressions, coverage, publication checks and resource cleanup. These compatible changes do not reset the observation start.
Native 0.9.1 adds cancellation observers. Its fixed-runner overhead and regression checks remain pending; the previous +3.70% measurement predates this implementation. The cancellation decision preserves short diagnostic samples and resource costs. The sporadic saturated CPU deadline investigation also remains open. These limitations and the elapsed-time gate prevent calling the whole roadmap complete.
Before 1.0, review the public API and dependency/default changes since the candidate, main/release CI, Native lifecycle and retained performance evidence, and any reported critical defects. External beta operation was canceled by the owner and is not a release gate. The documented Async limitation remains outside stable scope.