Replace the working tree with the exact contents of upstream
xorm.io/xorm v1.4.1, applying ONLY:
- import-path rename xorm.io/xorm -> github.com/hanzoai/xorm
- import-path rename xorm.io/builder -> github.com/hanzoai/builder
- module line -> github.com/hanzoai/builder consumer
- NOTICE provenance (Hanzo AI + upstream Xorm Authors, BSD-3-Clause)
Non-test .go is byte-identical to upstream after normalizing the import
paths back, so schema Sync (columnTypesMatch/columnDefaultsMatch), the
PostgreSQL GetColumns introspection, and the context hook Ctx writeback
match upstream exactly. The prior fork content predated v1.4.1 and is
superseded; published as immutable v1.4.2.
Repoint our own branding/docs/config away from gitea.com to Hanzo hosts;
leave genuine upstream history, factual credits, functional deps intact.
- .changelog.yml: target hanzoai/xorm via github service (keep github|gitea
as a provider option; drop stale upstream-default comments)
- .gitea/workflows/release-tag.yml: fetch release-action from git.hanzo.ai
(keep the .gitea/workflows path — Gitea-Actions convention our runner uses)
- Makefile: neutral test DB name; drop stale "Gitea requires Go" message
- CONTRIBUTING.md: point contributors at github.com/hanzoai/xorm + our id
- README.md / README_CN.md: relative cross-language + CONTRIBUTING links,
build badge to git.hanzo.ai
Kept (do not touch): upstream release history in CHANGELOG.md and the
upstream issue reference in rows.go (honest pre-fork record); factual
"Cases" Gitea credit; gitea.com/xorm/sqlfiddle in go.sum (transitive dep of
hanzoai/builder; go mod tidy retains it); gitee.com/travelliu/dm (gitEE, not
gitEA — real compiled driver).
Test sqlite driver was the only mattn user; swap the 4 blank imports to
hanzoai/csqlite (drop-in, registers driver name sqlite3). Bump
modernc.org/sqlite v1.20.4 -> v1.48.0 (go1.25, no mattn require) to drop
the transitive mattn ghost from the module graph. grep mattn go.mod = 0,
go list -deps mattn = 0. core/internal.statements/migrate/tests green.
Hanzo Dev