Files
iam/hanzo.yml
zeekayandHanzo Dev 1f31177689 ci: canonical pair — hanzo.yml + forge caller
github.com has no runner for hanzo-build-linux-amd64, so a caller under
.github/workflows is a gate that can never be scheduled. This is the ~7-line
caller on the plane that can (git.hanzo.ai git-runner fleet), pinned @v1.

Co-authored-by: Hanzo Dev <[email protected]>
2026-08-06 02:27:24 -07:00

31 lines
1.6 KiB
YAML

# Canonical CI/CD config for hanzoai/iam — the one file both the hanzoai/ci
# reusable (.hanzo/workflows/cicd.yml) and platform.hanzo.ai read.
#
# GATE ONLY. The image already has exactly one builder and it is deliberate:
# .hanzo/workflows/image.yml, which tags by COMMIT SHA (a semver tag that gets
# re-pushed leaves two digests behind one name, which is how platform's v4.4.5
# came to mean two builds) and mounts the token `go mod download` needs for the
# private hanzoai/orm + hanzoai/sqlite modules. That file's own header documents
# why it is the ONLY file in this repo that builds an image. Declaring `images:`
# here would make a second one, which is the exact failure it was written to end.
#
# The gate is the repo's own: `make test`. Two halves, both real —
# * zipdoc -check in every directory that generates one: a codegen-freshness
# refusal, so a handler doc comment that no longer matches its generated
# zipdoc_gen.go fails the build instead of drifting silently.
# * `go test ./... -race -count=1`: the whole suite (140 test files), under the
# race detector, with caching off so a green means it ran here and now.
# `go build ./...` runs first so a plain compile break fails in seconds rather
# than after the full race build.
#
# Note on what is NOT here: no `-tags skipCi`. Files guarded `//go:build !skipCi`
# vanish under that tag and `go test` then reports "[no tests to run]" and exits
# 0 — a green over zero tests. This tree carries no such guard and this gate
# passes no such tag; both halves of that have to stay true.
test:
- name: build
run: |
set -e
go build ./...
make test