From 87b3499164d30a6665e9b59a8606984e925ccd36 Mon Sep 17 00:00:00 2001 From: tannevaled Date: Mon, 12 Oct 2026 09:22:46 +0200 Subject: [PATCH 1/2] landing: the conformance figure and the gate it describes, both corrected The first page anybody reads said rbgo passes 22,488 ruby/spec examples, in four places. The gate's own file sums to 23,511. The second error is the more interesting one. Two of those places described CI as gating on "a shrink-only floor of 22,475" -- a mechanism that stopped existing when ruby#739 replaced it with a per-file baseline. So the page UNDERSOLD the guarantee as well as misstating the number: a scalar floor needed a margin wider than the largest spec file and so could not see a regression smaller than that, which is exactly why it was replaced. The page now says what CI does -- no spec file may pass fewer examples than its recorded row, and none may stop loading. WHAT WAS LEFT ALONE. The dated block in layouts/_default/list.html keeps its 22,488 and its 93.3% / 24,116 / 477 breakdown: it says "measured 2026-09-26 on main (3e8e3cc)" and a dated measurement going stale is not an error. Only its MECHANISM bullets changed, because how CI gates is not a measurement and was simply wrong. Verified unrelated and correct while here: "178 go-ruby-* modules" matches the count of go-ruby-* requires in rbgo's go.mod exactly. ADDED: the deploy workflow now runs the docs site's conformance check before hugo, so a wrong figure fails the build instead of being published, on pull requests as well as on main. It is consumed across repositories rather than copied -- two readers of one file cannot drift, where two transcriptions of one number already had -- and it needed an extension to see this page at all, since the landing style puts the figure before the words (go-embedded-ruby/docs#24). Co-Authored-By: Claude Opus 5 --- .github/workflows/deploy-pages.yml | 21 +++++++++++++++++++++ layouts/_default/list.html | 4 ++-- layouts/index.html | 6 +++--- 3 files changed, 26 insertions(+), 5 deletions(-) diff --git a/.github/workflows/deploy-pages.yml b/.github/workflows/deploy-pages.yml index b17ce26..9780b37 100644 --- a/.github/workflows/deploy-pages.yml +++ b/.github/workflows/deploy-pages.yml @@ -46,6 +46,27 @@ jobs: id: pages uses: actions/configure-pages@v6 + # This page states rbgo's ruby/spec total in the present tense in four + # places, and on 2026-10-12 all of them said 22,488 while the gate's own + # file summed to 23,511 — and two of them described the gate as a + # "shrink-only floor of 22,475", a mechanism replaced by a per-file + # baseline in ruby#739. A number and a mechanism, both stale, on the first + # page anybody reads. + # + # The check sums the gate's own per-file baseline and compares it against + # every present-tense claim here. It is the same binary the docs site + # runs, consumed across repositories rather than copied: two readers of + # one file cannot drift, where two transcriptions of one number already + # had. A dated block ("measured ") is exempt, because a dated + # measurement going stale is not an error. + # + # It runs BEFORE hugo so a wrong figure fails the build instead of being + # published, and on pull requests as well as on main. An unreachable + # baseline exits 2 and fails the step: an unreadable baseline finds no + # disagreement either, and that silence would read as agreement. + - name: The stated conformance total matches rbgo's gate + run: go run github.com/go-embedded-ruby/docs/tools/conformancefigure@main -root layouts + - name: Build landing page run: hugo --minify --baseURL "${{ steps.pages.outputs.base_url }}/" diff --git a/layouts/_default/list.html b/layouts/_default/list.html index 1153651..91f6127 100644 --- a/layouts/_default/list.html +++ b/layouts/_default/list.html @@ -70,10 +70,10 @@

C · Application conformance & performance

Runtime conformance — ruby/spec

22,488 examples pass in ruby/spec's language/ + core/ suites, measured 2026-09-26 on main (3e8e3cc). Three independent runs — two through the ratchet at different timeouts, one through a separate sweep — reported the same total.

    -
  • CI gates on a shrink-only floor of 22,475. The floor is the level a run may not fall below — it is not the score: it is what CI guarantees, while the measured total is what one run did.
  • +
  • CI gates file by file, not on a total: no spec file may pass fewer examples than its recorded row, and none may stop loading. The sum of those rows is 23,511 at rbgo v0.3.0 (2026-10-12) — it is what CI guarantees, while the measured total above is what one dated run did. This replaced a single shrink-only floor of 22,475, which needed a margin wider than the largest spec file and so could not see a regression smaller than that.
  • Only language/ and core/ are swept. ruby/spec's library/, optional/ and security/ trees are not run at all, so this figure is not a percentage of Ruby, and most of the standard library sits outside it.
  • Of the examples that ran, 93.3% pass (22,488 of 24,116); 477 more are skipped because our shim stubs the matcher they need, and 10 of 2,206 spec files produce no result. Skips and unreadable files sit outside the ratio entirely.
  • -
  • The floor has climbed from 6,000 on 2026-08-03 to 22,475. Two earlier jumps were fixes to the measurement — the mspec shim, and a parser upgrade that made 13 previously-unparseable spec files visible — which are gains in what the measurement can see, not in what the VM can do.
  • +
  • The guarantee has climbed from a floor of 6,000 on 2026-08-03 to a per-file baseline summing 23,511. Two earlier jumps were fixes to the measurement — the mspec shim, and a parser upgrade that made 13 previously-unparseable spec files visible — which are gains in what the measurement can see, not in what the VM can do.
diff --git a/layouts/index.html b/layouts/index.html index 3268400..a3e5f30 100644 --- a/layouts/index.html +++ b/layouts/index.html @@ -24,7 +24,7 @@

A Ruby you can embed,
written in Go.{{ partial "icons/rocket.html" . }}Get started → {{ partial "icons/graph.html" . }}See the benchmarks

-
22,488 ruby/spec language/+core/ examples passing~100% of real Ruby parsed6 64-bit arches
+
23,511 ruby/spec language/+core/ examples passing~100% of real Ruby parsed6 64-bit arches
rbgo — zsh
@@ -59,7 +59,7 @@

Everything a Ruby runtime gives you — as a Go library.

🧩

Embeddable

A real object model — classes, mixins, method_missing, metaprogramming, Fiber. The Go API is deliberately small today: ruby.Run(src, out) runs a program on a fresh VM and writes its output to one writer. Calling a Ruby method from Go is not exposed yet.

📦

One static binary

CGO=0 by default. Cross-compiles to amd64, arm64, riscv64, loong64, ppc64le and s390x.

⚡

AOT for integer kernels

Ahead-of-time compilation of integer-bound methods to unboxed Go. Measured on this project’s own micro-benchmarks (darwin/arm64, whole-process wall time): 8.7× MRI+YJIT on the tight-loop case, 2.0× on recursive fib(35). Only methods the compiler can lower benefit.

-
✅

Conformance, measured

22,488 ruby/spec language/ + core/ examples pass, gated in CI by a shrink-only floor of 22,475 — the floor is what CI refuses to fall below, not the score. Only language/ and core/ are swept: library/, optional/ and security/ are not run, so this is not a percentage of Ruby. Front-end: ~100% of valid Ruby parsed across Rails + Puppet (2026-06).

+
✅

Conformance, measured

23,511 ruby/spec language/ + core/ examples pass, gated in CI file by file: no spec file may pass fewer examples than its recorded row, so a regression is named rather than absorbed in a total. Only language/ and core/ are swept: library/, optional/ and security/ are not run, so this is not a percentage of Ruby. Front-end: ~100% of valid Ruby parsed across Rails + Puppet (2026-06).

🔒

Closed-world builds

rbgo build --closed bakes the program to bytecode and drops the front-end for a smaller binary.

🌿

A real ecosystem

Built from the go-ruby-* family — each a 100%-coverage, 6-arch stdlib module, bound natively.

@@ -68,7 +68,7 @@

Everything a Ruby runtime gives you — as a Go library.

8.7×
rbgo+AOT vs MRI+YJIT, tight-loop benchmark (2.0× on fib(35))
-
22,488
ruby/spec language/+core/ examples passing · CI floor 22,475
+
23,511
ruby/spec language/+core/ examples passing · gated per file, not by a total
178
go-ruby-* modules bound into rbgo
CGO=0
across 6 arches · 3 OSes
From 2428230eb48ea650e7b22a9bcaaa486b9d280292 Mon Sep 17 00:00:00 2001 From: tannevaled Date: Mon, 12 Oct 2026 09:28:02 +0200 Subject: [PATCH 2/2] landing: pin the conformance check to a tag, because @main is not the current main MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The check failed on this branch while its figures were right. The tool it ran was the proxy's cached @main, nine minutes stale, and that version cannot see this page's phrasing -- so it reported "0 current claim(s) checked" and exited 1, which is the correct behaviour for an unmatched vocabulary and the wrong answer for this page. Measured, because "the proxy lags" is the kind of claim that deserves it: $ go list -m github.com/go-embedded-ruby/docs@main github.com/go-embedded-ruby/docs v0.0.0-20261012071659-5a27ddae9f8d 07:16:59 $ gh api repos/go-embedded-ruby/docs/commits/main --jq .sha 3384011416dc… 07:25:48 $ GOPROXY=direct go list -m github.com/go-embedded-ruby/docs@main github.com/go-embedded-ruby/docs v0.0.0-20261012072548-3384011416dc correct So `@main` does not mean "the current main"; it means whatever the proxy last cached for that branch. Pinned to go-embedded-ruby/docs v0.1.0, which resolves deterministically and passes here: 4 current claim(s) checked against the gate's 23511. Same family as the local toolchain lying about its own version: a name that looks like it designates a version can designate a cache instead, and only asking resolves it. Co-Authored-By: Claude Opus 5 --- .github/workflows/deploy-pages.yml | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-) diff --git a/.github/workflows/deploy-pages.yml b/.github/workflows/deploy-pages.yml index 9780b37..428f414 100644 --- a/.github/workflows/deploy-pages.yml +++ b/.github/workflows/deploy-pages.yml @@ -64,8 +64,21 @@ jobs: # published, and on pull requests as well as on main. An unreachable # baseline exits 2 and fails the step: an unreadable baseline finds no # disagreement either, and that silence would read as agreement. + # + # PINNED TO A TAG, not @main, and that is not tidiness. `@main` is not + # "the current main": it is whatever the module proxy last cached for the + # branch, and the lag is measurable -- + # + # $ go list -m github.com/go-embedded-ruby/docs@main + # …v0.0.0-20261012071659-5a27ddae9f8d # 07:16:59 + # $ gh api …/commits/main --jq .sha + # 3384011416dc… # 07:25:48 + # + # The first version of this step used @main and failed on a branch whose + # figures were right, because the tool it ran was nine minutes old and + # could not yet see this page's phrasing. A tag resolves deterministically. - name: The stated conformance total matches rbgo's gate - run: go run github.com/go-embedded-ruby/docs/tools/conformancefigure@main -root layouts + run: go run github.com/go-embedded-ruby/docs/tools/conformancefigure@v0.1.0 -root layouts - name: Build landing page run: hugo --minify --baseURL "${{ steps.pages.outputs.base_url }}/"