Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions site/announcements/index.html
Original file line number Diff line number Diff line change
Expand Up @@ -14,6 +14,7 @@
<h1>Announcements</h1>
<p>Press releases and release announcements for sc-compose.</p>
<ul>
<li><a href="sc-compose-v1.6.1.md">sc-compose v1.6.1</a> — ARM64 Linux Support, Go-Native Packaging, Beads Runner Reliability</li>
<li><a href="sc-compose-v1.6.0.md">sc-compose v1.6.0</a> — Beads Composition Engine, Renderer Hardening, Release Consolidation</li>
<li><a href="sc-compose-v1.5.0.md">sc-compose v1.5.0</a> — Hermes Runtime, sc-publish migration, Go distribution, release hardening</li>
</ul>
Expand Down
49 changes: 49 additions & 0 deletions site/announcements/sc-compose-v1.6.1.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,49 @@
# sc-compose v1.6.1 — ARM64 Linux Support and Beads Runner Reliability

**Released:** August 30, 2026 · **Install:** `cargo install sc-compose` (Rust), `pip install sc-compose` (Python ≥3.11), `brew install randlee/tap/sc-compose` (macOS)

[Changelog](https://github.com/randlee/sc-compose/blob/main/CHANGELOG.md) · [Release notes](https://github.com/randlee/sc-compose/releases/tag/v1.6.1)

---

## Python Developer

**As a Python developer, I want Python bindings and pytest-fixture generation, so that I can drive composition from Python and generate test stubs from a list of test names.**

v1.6.1 extends the pre-built wheel matrix with ARM64 Linux. `pip install sc-compose` now serves `manylinux_2_17_aarch64` wheels alongside the existing macOS, x86_64 Linux, and Windows targets, so Python developers on ARM64 Linux hosts — AWS Graviton, Ampere, and Linux VMs on Apple Silicon — get a compiled native extension straight from PyPI instead of building from source through maturin.

The same ARM64 Linux coverage flows through the whole distribution family: the `sc_composer_beads` and `sc_sha` wheels gain matching `aarch64` builds, keeping the Python surface uniform across architectures.

---

## Release & Ops Engineer

**As a release/ops engineer, I want to compose service configs from shared fragments and generate compliance/sprint reports, so that configs stay DRY and reports are reproducible.**

The release target matrix gains ARM64 Linux, shipping a native `aarch64-unknown-linux-gnu` binary alongside the existing macOS, Windows, and x86_64 Linux artifacts. The release pipeline itself also takes a meaningful step: sc-compose now adopts the generated `sc-sha-go` native-module packaging for its own build/release flow (Sprint S.11), so the pipeline that produces the binary is built on the same reproducible native-library layout it publishes to downstream Go consumers.

CI reliability gets tightened in the same pass: the `sprint/*` branch trigger cascade is fixed so stacked sprint PRs fire CI correctly, and the ARM64-dependent lint and release-validation jobs are gated and isolated so external lint integration can't stall or fail unrelated targets.

---

## Beads Alchemist

**As a beads alchemist, I want to compose and instantiate beads formulas — workflow, expansion, aspect, and convoy — through sc-compose, so that I can express a recurring multi-step agent workflow once and pour it across projects.**

The Beads formula runner gets a reliability pass (Sprint S.8): capture lifecycle is simplified, output-reader failures are contained so a stray capture error can't take down the whole run, a process-tree timing race in the runner's tests is removed, and the capture reader join is now proven by dedicated tests rather than assumed. No new formula surface ships — this is hardening, not feature work.

---

## Library Consumer (Rust)

**As a Rust library consumer, I want the `sc-composer` crate with a stable API, so that I can embed composition in my own tooling.**

Phase S hotspot remediation lands internally: complexity is reduced and boundary invariants are hardened across the extractor and template-lint seams, with zero public-API or behavior change. The `sc-composer` crate surface stays exactly as it was in v1.6.0 while the internals get measurably more maintainable. The arm64 Linux native-module build also extends the targets available to the generated `sc-sha-go` module, keeping the Go-consumer native library in lockstep with the new release matrix.

> The AI-Agent Profile Author and Task-Template Author personas are unchanged this release — the extractor and template-lint refactoring above is internal-only (no behavior change), and no new profile, task-template, or rendering surface ships in v1.6.1.

---

## What's Next

With ARM64 Linux delivery landed and the Go-native packaging adopted for sc-compose's own pipeline, subsequent work continues the Phase S hardening loop and the Beads runner reliability cadence established here.
Loading