Faber 1.1.0

Install this version#

No prebuilt archives were published for this version. It is listed here because its release notes are part of the record.

Release notes#

Minor product release spanning 13 commits since v1.0.0 (2026-07-14→2026-07-17). Headline: reader-locale-driven Faber emit surface (--reader-locale), the first tag-driven GitHub Actions release workflow, and the process/versioning/interdependency analysis docs.

*Era note: this tag predates the current release-note convention (Scale table, companion pins, recorded verification gates). These notes are reconstructed post-hoc from the commit range; no pre-release gate records were captured in-tree for this era.*

Scale#

Remainder of the range: 4 docs, 1 ci, 1 release, 1 polish.

git log v1.0.0..v1.1.0 --oneline --no-merges

Major tracks#

Reader-locale emit surface (--reader-locale)#

Phase 2 of the reader-locale-emit goal (Radix). The two CLI gates that suppressed localized output are removed: faber emit -t faber no longer rejects --reader-locale, and faber format no longer requires --canonical. The reader locale resolves to a pack (Faber owns install-layout + package-manifest resolution) and is handed to radix::tool::cmd_emit_with_reader_pack, so `faber emit -t faber --reader-locale=<X> emits localized Faber. --canonical` remains the Latin alias; no flags keeps author mode.

Phase 3 proves the surface: a new test shows --reader-locale=ar emits Arabic keywords (دالة functio، بداية incipit) in logical codepoint order — no Latin survives and no Bidi isolates / embedding controls (U+202A–U+202E, U+2066–U+2069) are injected.

Go codegen dispatch adaptation#

The radix generate_from_analyzed dispatch seam gained a KeywordSurface parameter after reader-locale emit threading. Go is not a localized target, so the surface is unused; a Latin surface is constructed once per Go package result and forwarded to keep the build green.

First tag-driven release workflow#

.github/workflows/release.yml lands (b771357): triggers on vX.Y.Z tag push or manual dispatch, builds the faber binary for x86_64-linux, x86_64-macos, and aarch64-macos, validates the crate version against the tag, and publishes platform archives with SHA-256 checksums to faberlang/releases with faber-prefixed tags. The CI workspace assembles the full faberlang/ layout by checking out siblings (radix via FABERLANG_RELEASES_TOKEN, cista, faber-runtime, host-kernel-rs, host-native-rs, host-providers-rs). The hygiene ratchet is vendored as crates/hygiene-ratchet/ so Faber is self-contained for hygiene tests (the cross-repo path dep was a monorepo split leftover).

Package resilience#

fix(package): fall back from a missing runtime dependency instead of failing, with new test coverage. A generated-provider loopback test (test(http)) proves the provider codegen round-trip. Artifact-plan helpers are clarified.

Process docs#

• docs(design): CLI surface analysis vs Radix + Cista (head-cto) — the compile/check/run/test loop is fully covered (14 targets); package management is flagged as the main usability gap (by-design repo separation, staged per the Cista roadmap); no gap blocks the release-1.1 narrative.
• docs(release): process, versioning, and interdependency analysis (head-cpo) across all 10 repos, three release workflows, and version/tag drift.
• docs(release): major parity and language lock definition.
• docs: faber factory status truth refresh.

Version alignment#

Known limitations#

• The tag-driven release workflow landed the same day the v1.1.0 tag was cut; CI cargo build --locked against sibling default-branch tips was not yet reconciled. The same-day v1.1.1 re-release refreshes Cargo.lock specifically so the workflow's locked build succeeds (see v1.1.1.md).
• No faber-v1.1.0 public binary was published to faberlang/releases.
• No in-tree pre-release verification records exist for this era.

---

All releases · Install the current release