INT-10: affinescript-lsp distribution (consume the ADR-019 compiler shim)

[hyperpolymath]

Ledger INT-10 (was blocked by INT-04). Unblocked by ADR-019 / #260 (compiler distribution: GitHub Releases binaries + thin Deno/JSR shim @hyperpolymath/affinescript).

Scope

affinescript-lsp is distributed/installed such that it resolves a working compiler via the ADR-019 shim (deno add @hyperpolymath/affinescript → shim downloads+verifies the pinned Release binary). No bespoke compiler-bundling in the LSP.

Sequencing

• Blocked on INT-04: compiler distribution strategy (native OCaml binary — not JSR/npm) #260 S2 (release.yml per-platform binary + SHA256SUMS) and S3 (the shim package).
• Then S4 of INT-04: compiler distribution strategy (native OCaml binary — not JSR/npm) #260 == this issue: wire affinescript-lsp onto the shim + a smoke that the LSP can locate/exec the compiler.

Refs #260, #181. ADR-019 in docs/specs/SETTLED-DECISIONS.adoc + .machine_readable/6a2/META.a2ml.

[hyperpolymath]

Reopened: this was auto-closed when PR #285 (#260 S3 shim) was merged via the gitbot's Refs-auto-close behaviour. The actual INT-10 wiring (affinescript-lsp consuming the @hyperpolymath/affinescript shim) is not yet done — gated on the v0.1.0 Release workflow (run 26124042846, still queued) producing assets so pins.js can be filled and the shim version bumped (#260 S4). Restoring the true state.

[hyperpolymath]

Update — diagnosed why the v0.1.0 Release workflow's macos-x64 leg has been queued ~10 h: it's not CI concurrency, it's that macos-13 GitHub-hosted runners were fully retired on 2025-12-04 (actions/runner-images#13046, #13402, #13634). The release.yml matrix was pinning a runner image that no longer exists; the job was never going to start.

Actions taken just now:

1. Cancelled run 26124042846 (cancellation only — release assets are intact).
2. Opened ci(release): swap retired macos-13 runner for macos-15-intel #292: swap os: macos-13 → os: macos-15-intel (the current Intel x64 standard runner per the actions/runner-images README; stable GA, in contrast to the still-stabilising macos-26-intel).
3. PR feat(shim): #260 S4 — finalise pins.js for v0.1.0 (partial), bump shim 0.1.1 #291 stays open for the partial pins.js fill (linux-x64 + macos-arm64); it leaves macos-x64 fail-closed as documented.

Once #292 merges and the next v* tag is cut, the workflow will produce the missing macos-x64 binary + a complete SHA256SUMS. The follow-up shim release fills the last pin in pins.js, and #282 can finally close.