docs(adr-0056): Scoop+Homebrew validated; record the Caddy HEAD fix

Amendment 3: scoop install and brew install both verified end-to-end on
real Windows + macOS (install + run), resolving Amendment 2's unverified
caveats (incl. the HEAD:main push).

Records the root cause + fix for the Homebrew 403: Gitea (>=1.25)
303-redirects release downloads to a method-bound SigV4 presigned S3 URL;
brew resolves with HEAD, captures the HEAD-signed URL, then GETs it -> 403
(GET-only tools are unaffected). Fixed by a server-side Caddy rule that
answers HEAD on release-download paths directly (200, no redirect) so the
download GET re-runs the redirect fresh. That rule lives in the Gitea edge
config, NOT this repo - documented so a server rebuild doesn't silently
break brew. Includes the reference Caddyfile block.

Also notes the ad-hoc mac signature runs fine via brew; Developer-ID +
notarization stays parked on the Apple org conversion. Remaining D3: winget.

docs/adr/0056-crates-io-and-cargo-binstall.md

@@ -176,3 +176,65 @@ whether macOS Gatekeeper accepts the **ad-hoc-signed** mac binary via
 requirement and `brew`'s curl download sets no quarantine xattr, unlike a
 browser download — but this rides on the still-parked Developer-ID signing
 decision). **Remaining D3:** winget (komac on Linux CI, or a manual PR).
+
+## Amendment 3 — 2026-06-19: validated end-to-end; a presigned-URL/HEAD gotcha + the Caddy fix it needs
+
+**Scoop and Homebrew now install and run end-to-end** (`v0.2.0`,
+user-verified on real Windows + macOS): `scoop install rdbms-playground`
+and `brew install lazyeval/tap/rdbms-playground` both fetch, checksum,
+install, and the installed binary launches. Amendment 2's "unverified"
+caveats are resolved (the `HEAD:main` push populated both repos cleanly).
+
+**Distribution depends on a server-side Caddy rule — RECORD THIS: it lives
+in the Gitea edge config, NOT this repo, and if it is lost in a server
+rebuild Homebrew silently 403s again.** Symptom: `brew install` failed with
+`curl (22) … 403` on the asset URL. Root cause: Gitea (≥1.25; here 1.26.2)
+serves release-asset downloads by **303-redirecting to a method-bound
+AWS-SigV4 presigned S3 URL** (OVH), signed for the *incoming* request's HTTP
+verb. Homebrew **resolves the URL with a HEAD, captures the returned
+(HEAD-signed) presigned URL, then runs the download GET against that
+captured URL** → GET-on-a-HEAD-signed-URL → 403. Reproduced precisely:
+`GET on HEAD-resolved = 403`, `HEAD on HEAD-resolved = 200`, `GET on
+GET-resolved = 200`, `HEAD on GET-resolved = 403`. GET-only tools
+(`install.sh`, `install.ps1`, `cargo binstall`, curl) are unaffected — they
+GET the original URL and let it redirect fresh to a GET-signed URL. This is
+a **Homebrew defect** (it reuses an ephemeral, method-scoped credential as
+if it were a durable resource locator); Gitea's per-request method-bound
+signing is correct, and SigV4 *cannot* sign one URL for two verbs.
+`SERVE_DIRECT=false` (Gitea proxies, no presign) would also fix it but was
+declined — not reshaping storage for one client.
+
+**Fix (deployed, "Tier A"): a Caddy rule that answers HEAD on
+release-download paths directly** — 200, no redirect, no body — so brew's
+resolve records the *original* URL (no presigned credential captured) and
+its download GET runs through the redirect fresh → GET-signed → 200. Scoped
+to `method HEAD` + the release-download path; **GET is untouched**, so every
+working channel is unaffected, and a HEAD carries no payload so no client
+can *break* (at most a HEAD-probing download manager loses a progress-bar
+size — cosmetic). Reference Caddyfile (place before the Gitea
+reverse_proxy):
+
+```
+@release_head {
+    method HEAD
+    path_regexp ^/[^/]+/[^/]+/releases/download/.+
+}
+handle @release_head {
+    header Accept-Ranges bytes
+    header Content-Type application/octet-stream
+    respond 200
+}
+```
+
+A "Tier B" variant (a sidecar that fetches and returns the *real*
+Content-Length so even HEAD-probing clients stay fully faithful) was specced
+but proved unnecessary — brew is happy with the bare 200.
+
+**macOS signing:** the brew-installed binary **runs** under the current
+**ad-hoc** signature (`codesign --sign -`) — confirmed on Apple Silicon.
+**Developer-ID signing + notarization remains parked** pending the user's
+Apple account conversion to an Organization (GA plan / ADR-ci-003); it is
+needed for *browser-download* trust, not for the package-manager paths,
+which are all working now.
+
+**Remaining D3:** winget only.

docs/adr/README.md

@@ -68,4 +68,4 @@ This directory contains the project's ADRs, recorded per
 - [ADR-0053 — Contextual `hint` command and keybinding](0053-contextual-hint-command-and-keybinding.md) — **Accepted, implemented 2026-06-15** (Phases A–D; closes **A1** + requirements **H2**). Settles the `hint` slot ADR-0003 left "ADR pending"; closes the last open piece of **A1** and tracks requirements **H2**. **Two surfaces:** an **F1 keybinding** that renders a deep hint for the *live* partial input without submitting (the primary path — a submitted `hint` command can't see the buffer it would help with, since Enter empties it), and a submitted **`hint` command** that expands on the *most recent error*. **No topic argument** (contextual only — `help <topic>` already owns explicit reference). Introduces a **tier-3 teaching layer**, deeper than the existing tier-1 (colour / error headline) and tier-2 (ambient one-liner; and the error `hint:`, which is shown **by default** since `Verbosity::Verbose` is the default — `messages short` is the opt-*out*); without it `hint` would just duplicate what's already on screen. Tier-3 content lives in the catalogue under `hint.cmd.<hint_id>` (per command form) and `hint.err.<class>` (per error/diagnostic class), each a structured `what`/`example`/`concept` block rendered via a new `note_hint*` family with `OutputStyleClass::Hint`. **Keyed per-form via a new `hint_ids: &[&str]` field on `CommandNode` mirroring `usage_ids`** (revised in Phase B): a per-*node* key proved too coarse — `add`/`drop`/`show`/`create` are each one node spanning many forms, and a live-input hint for `add 1:n relationship` must be specific to relationships; `hint_key_for_input_in_mode` reuses `usage_key_for_input_in_mode`'s form-word disambiguation, and covers the advanced-SQL forms whose `usage_ids` are empty. Not keyed off `help_id` (it is `None` on the advanced-SQL nodes purely to dedup the `help` list; that parallel gap is issue **#36**). **Clause-concept hints** (`on delete` actions, constraint slots, `with pk`, cardinality) are a recorded **deferred extension** (`hint.concept.<topic>`, issue **#37**) — per-form is the right tier-3 granularity, with position-awareness owned by tier-2 + the live `Next:` line. Runtime `translate_error` classes resolve via stored `last_error_hint_key` (`hint` command / empty-F1). (The second route — pre-submit `diagnostic.*` read live from the walker on the F1 path — is **deferred**, issue **#38**: `Diagnostic` carries no class key.) Adds `AppCommand::Hint`, a `HINT` grammar node + REGISTRY entry, the `hint_ids` field, and `last_error_hint_key`; F1 is a read-only overlay (buffer + completion memo untouched). **Content is the bulk of the work** (the mechanism is ~a day): v1 scope = ~37 command forms + 9 runtime error classes (comprehensive for those, ~57 blocks), authored **exemplars-first** (voice approved in this ADR's `/runda` review, then mass-authored in batches), enforced by a **comprehensiveness coverage test**, with graceful fall-back to tier-2 if a key is ever missing. The **pre-submit-diagnostic route + ~33 `diagnostic.*` blocks were deferred** (issue **#38**) — `Diagnostic` carries no class key, so the route needs a broad change for marginal value (tier-2 already surfaces diagnostics; many duplicate runtime classes). Forks user-chosen: two-surface model; **F1** (vs `?` / a chord); no-arg; comprehensive-for-commands-and-errors scope; exemplars-first; diagnostics deferred. OOS: per-topic `hint <topic>` (rejected — overlaps `help`); always-on tier-3 (rejected — keeps ambient terse); non-`en-US` locales + success-command teaching (deferred); clause-concept hints (issue #37); the diagnostic route (issue #38); the `help`-side advanced-SQL gap (issue #36)
 - [ADR-0054 — Release versioning policy + version surfaces (`--version` / `version`)](0054-release-versioning-and-version-surfaces.md) — **Accepted + implemented 2026-06-16** (plan: `docs/plans/20260616-public-availability.md`, step 1 on the road to public availability; no prior issue/`requirements.md` item — an untracked gap). Fixes the **tag↔crate-version decoupling**: `Cargo.toml` built `0.1.0` while `release.yaml` named assets from the git tag, so a binary could report a version different from the asset it shipped in. **Decision:** `Cargo.toml` `version` is the **single source of truth** (read via `env!("CARGO_PKG_VERSION")`, no tag-injection); two surfaces report it through one `cli::version_text()` → catalog `cli.version_line` — a **`--version` / `-V`** CLI flag (mirrors `--help`, prints+exits in `main.rs`) and an in-app **`version`** command (REGISTRY node `app::VERSION`, `AppCommand::Version`, emits via `note_system`); and a **release-CI version guard** (`release.yaml` `test` job reads the `[package]` version from `Cargo.toml` and **fails the release** unless the `v*` tag equals `v<version>`; the guard's parse was later switched from `cargo metadata | node` to a `grep` on Cargo.toml after the former broke on the flake devShell's stdout banner). Release ritual: bump `Cargo.toml` → commit → tag → push. New keys `cli.version_line` + `help.app.version` + `parse.usage.version` + `hint.cmd.version.{what,example}` (the new REGISTRY command pulls in the comprehensiveness coverage gate). Rejected: tag-as-source (makes Cargo.toml lie). Deferred: git-hash/build-date enrichment (behind the same `version_text()` seam); UI placement beyond the command. Tested test-first: CLI parse (`--version`/`-V`/default-off), `version_text()` carries `CARGO_PKG_VERSION`, the in-app command parses + emits. Also corrected a stale `release.yaml` header comment ("macOS is deferred" → built by the dispatched `release-macos.yaml`).
 - [ADR-0055 — `curl | sh` install script (`scripts/install.sh`)](0055-curl-sh-install-script.md) — **Accepted + implemented 2026-06-17** (plan: `docs/plans/20260616-public-availability.md`, step 2; tracked by plan + ADR, no Gitea issue — user decision). A one-line installer (`curl -fsSL <gitea-raw>/scripts/install.sh | sh`) so beginners don't hand-pick an asset + `chmod +x`. **POSIX `sh`** (shellcheck-clean), detects `uname` OS/arch → target triple (**Linux → the fully-static `*-musl`** build, macOS → `*-apple-darwin`; `amd64`/`arm64` aliased; **Windows rejected** → Scoop/winget/releases page), resolves the version from the **`releases/latest`** API (or `RDBMS_VERSION` to pin), downloads the asset **and its `.sha256` and verifies it** (mismatch aborts), installs to `~/.local/bin` (`RDBMS_INSTALL_DIR` override) with a PATH hint. Testing seams: `RDBMS_OS`/`RDBMS_ARCH` + `--print-target`. macOS note: `curl` downloads aren't Gatekeeper-quarantined so the ad-hoc binary runs as-is (Developer-ID + notarization is the postponed signing task). **Verified end-to-end against the live public `v0.1.0`** (all platform mappings, pinned + latest, checksum incl. tamper-rejection, install + run). Rejected: website-domain hosting (extra moving part; Gitea raw is simplest); deferred: uploading the script as a release asset, and a **shellcheck CI gate** (shellcheck isn't in the flake — touches ADR-ci-002). **Amendment 1 (2026-06-17):** added a Windows **`scripts/install.ps1`** (`irm | iex`; maps host CPU → our `*-windows-gnu`/`-gnullvm` `.exe`, SHA-256-verifies, installs to `%LOCALAPPDATA%\Programs\…` + user PATH) — user chose both a one-liner *and* Scoop/winget; **written but untested from this env** (no PowerShell — validate on Windows).
-- [ADR-0056 — crates.io publish-readiness + `cargo binstall` metadata (D3)](0056-crates-io-and-cargo-binstall.md) — **Prepared 2026-06-17** (plan step 3a; tracked by plan + ADR). Makes the crate **ready to publish** to crates.io (user decision) and adds `cargo-binstall` metadata; the actual `cargo publish` is a **gated, irreversible maintainer step**. Manifest: drops `publish = false`; adds `homepage` (relplay.org), `keywords`, `categories`, and an `exclude` (`/website`,`/docs`,`/.gitea`,`/.codegraph`) trimming the crate from 585 files/8.3 MiB → **353/913 KiB compressed** (code-only). Authors **`README.md`** (engine-neutral, simple/advanced-mode wording; install via curl|sh/binstall/source/prebuilt) and **`LICENSE-MIT`** (© Lazy Evaluation Ltd — *confirm holder*); the canonical **`LICENSE-APACHE`** is deferred to the maintainer (don't ship retyped legal text) — the SPDX `license` field already satisfies crates.io. **binstall** (syntax verified vs cargo-binstall SUPPORT.md): `pkg-fmt = "bin"` (bare binaries), `pkg-url` spelled `v{ version }` (the placeholder omits the `v`), plus per-target **`overrides`** mapping the common host triples to the assets we ship — `*-linux-gnu` → the static `*-linux-musl` build, `*-pc-windows-msvc` → `*-gnu`/`-gnullvm` `.exe` (macOS matches directly; the docs promise no automatic fallback). **Ordering:** publish at a **new tagged version whose release exists**, after the release — **not `0.1.0`** (diverges from the already-released 0.1.0 binaries that predate `--version`). Verified: `cargo publish --dry-run` packages + verify-builds; `cargo metadata` confirms the binstall block + 4 overrides. **Unverified:** a real `cargo binstall` run (not a dep; nothing on crates.io yet) — validate at first publish. Rejected: cargo-dist (GitHub-centric). Maintainer follow-ups: confirm © holder, add canonical `LICENSE-APACHE`, real binstall validation. **Amendment 1 (2026-06-18):** `0.2.0` **published live** (crates.io; `cargo install` + `cargo binstall` verified — the unverified-overrides caveat is resolved), via a new **manual `workflow_dispatch`** workflow `.gitea/workflows/publish.yaml` (mirrors `release-macos.yaml`; `tag` input; `cargo publish` with a crate-scoped `CARGO_REGISTRY_TOKEN` secret). Publish stays **manual** by decision — irreversible (keeps the token off every tag push), the split release (tag Linux/Windows + dispatched macOS) makes a human the "all assets up" gate, and crates.io has no Gitea-Actions trusted-publishing path. Each registry is its **own idempotent job** (crates.io job no-ops if the version exists) so Scoop/Homebrew/winget can be added as sibling jobs without interfering. **Amendment 2 (2026-06-19):** **Scoop + Homebrew wired** (D3 §3b/§3c) as sibling `publish.yaml` jobs (`scoop-bucket`, `homebrew-tap`) that render manifests from the release `.sha256` sidecars and push to **org-level, multi-package** repos `lazyeval/scoop-bucket` + `lazyeval/homebrew-tap`. Credential: a scoped bot user **`lazyeval-ci`** (Gitea PATs scope by permission-category, not per-repo, so an `oli` token would over-reach to the main repo) on a `lazyeval` org team with Write to the package repos only; its PAT is the `LAZYEVAL_PKG_TOKEN` secret on `oli/rdbms-playground`. Render scripts (`scripts/render-{scoop-manifest,homebrew-formula}.sh`) are **dependency-free bash** (CI image `node:22-slim` has no jq/ruby), tested by `scripts/test-package-renders.sh`. Scoop: `#/`-rename fragment + `checkver`, no `autoupdate`. Homebrew: `on_macos`/`on_linux`×arch bare-binary formula, no Windows. **Unverified:** real `scoop`/`brew install`, the `HEAD:main` branch assumption, macOS Gatekeeper-via-brew (ad-hoc sign). **Remaining D3:** winget.
+- [ADR-0056 — crates.io publish-readiness + `cargo binstall` metadata (D3)](0056-crates-io-and-cargo-binstall.md) — **Prepared 2026-06-17** (plan step 3a; tracked by plan + ADR). Makes the crate **ready to publish** to crates.io (user decision) and adds `cargo-binstall` metadata; the actual `cargo publish` is a **gated, irreversible maintainer step**. Manifest: drops `publish = false`; adds `homepage` (relplay.org), `keywords`, `categories`, and an `exclude` (`/website`,`/docs`,`/.gitea`,`/.codegraph`) trimming the crate from 585 files/8.3 MiB → **353/913 KiB compressed** (code-only). Authors **`README.md`** (engine-neutral, simple/advanced-mode wording; install via curl|sh/binstall/source/prebuilt) and **`LICENSE-MIT`** (© Lazy Evaluation Ltd — *confirm holder*); the canonical **`LICENSE-APACHE`** is deferred to the maintainer (don't ship retyped legal text) — the SPDX `license` field already satisfies crates.io. **binstall** (syntax verified vs cargo-binstall SUPPORT.md): `pkg-fmt = "bin"` (bare binaries), `pkg-url` spelled `v{ version }` (the placeholder omits the `v`), plus per-target **`overrides`** mapping the common host triples to the assets we ship — `*-linux-gnu` → the static `*-linux-musl` build, `*-pc-windows-msvc` → `*-gnu`/`-gnullvm` `.exe` (macOS matches directly; the docs promise no automatic fallback). **Ordering:** publish at a **new tagged version whose release exists**, after the release — **not `0.1.0`** (diverges from the already-released 0.1.0 binaries that predate `--version`). Verified: `cargo publish --dry-run` packages + verify-builds; `cargo metadata` confirms the binstall block + 4 overrides. **Unverified:** a real `cargo binstall` run (not a dep; nothing on crates.io yet) — validate at first publish. Rejected: cargo-dist (GitHub-centric). Maintainer follow-ups: confirm © holder, add canonical `LICENSE-APACHE`, real binstall validation. **Amendment 1 (2026-06-18):** `0.2.0` **published live** (crates.io; `cargo install` + `cargo binstall` verified — the unverified-overrides caveat is resolved), via a new **manual `workflow_dispatch`** workflow `.gitea/workflows/publish.yaml` (mirrors `release-macos.yaml`; `tag` input; `cargo publish` with a crate-scoped `CARGO_REGISTRY_TOKEN` secret). Publish stays **manual** by decision — irreversible (keeps the token off every tag push), the split release (tag Linux/Windows + dispatched macOS) makes a human the "all assets up" gate, and crates.io has no Gitea-Actions trusted-publishing path. Each registry is its **own idempotent job** (crates.io job no-ops if the version exists) so Scoop/Homebrew/winget can be added as sibling jobs without interfering. **Amendment 2 (2026-06-19):** **Scoop + Homebrew wired** (D3 §3b/§3c) as sibling `publish.yaml` jobs (`scoop-bucket`, `homebrew-tap`) that render manifests from the release `.sha256` sidecars and push to **org-level, multi-package** repos `lazyeval/scoop-bucket` + `lazyeval/homebrew-tap`. Credential: a scoped bot user **`lazyeval-ci`** (Gitea PATs scope by permission-category, not per-repo, so an `oli` token would over-reach to the main repo) on a `lazyeval` org team with Write to the package repos only; its PAT is the `LAZYEVAL_PKG_TOKEN` secret on `oli/rdbms-playground`. Render scripts (`scripts/render-{scoop-manifest,homebrew-formula}.sh`) are **dependency-free bash** (CI image `node:22-slim` has no jq/ruby), tested by `scripts/test-package-renders.sh`. Scoop: `#/`-rename fragment + `checkver`, no `autoupdate`. Homebrew: `on_macos`/`on_linux`×arch bare-binary formula, no Windows. **Unverified:** real `scoop`/`brew install`, the `HEAD:main` branch assumption, macOS Gatekeeper-via-brew (ad-hoc sign). **Remaining D3:** winget. **Amendment 3 (2026-06-19):** **Scoop + Homebrew validated end-to-end** on real Windows + macOS (install + run). Found + fixed a **presigned-URL/HEAD gotcha**: Gitea (≥1.25) 303-redirects release downloads to a **method-bound SigV4 presigned S3 URL**; Homebrew resolves via HEAD, captures the HEAD-signed URL, then GETs it → 403 (GET-only tools unaffected). Fix is a **server-side Caddy rule** (lives in the Gitea edge config, NOT this repo — record it: lost on a rebuild → brew 403s again) answering HEAD on `…/releases/download/…` directly (200, no redirect) so brew's download GET re-runs the redirect fresh. ad-hoc mac signature **runs** via brew; Developer-ID + notarization still parked on the Apple org conversion. **Remaining D3:** winget only.