CI test harness

.github/workflows/test.yml

@@ -0,0 +1,23 @@
+name: test
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      # Python, Node.js, Perl, GCC/G++ 14 come preinstalled on the runner.
+      - name: Install toolchains not preinstalled on the runner
+        run: |
+          sudo apt-get update
+          sudo apt-get install -y --no-install-recommends racket
+
+      - uses: oven-sh/setup-bun@v2
+
+      - name: Run all language tests
+        run: ./run-tests.sh

ROADMAP.md

@@ -0,0 +1,67 @@
+# Roadmap
+
+Working agreement: every step is sized to be completed (and verified) in one
+sitting. An implementation only gets an `expected-output.txt` captured from an
+actual observed run — never written from reading the source. Anything that
+can't be run locally gets verified on the CI runner before merge.
+
+## Milestone 1 — CI backfill: every existing language tested on every PR
+
+The harness: `run-tests.sh` loops over `languages/*/*/test.sh`, runs each, and
+diffs stdout against that folder's `expected-output.txt`. The GitHub Actions
+workflow (`.github/workflows/test.yml`) runs it on every PR and push to main.
+Folders without a `test.sh` are skipped until backfilled.
+
+- [x] **Step 1 — harness + locally verifiable languages.** Workflow file,
+      `run-tests.sh`, and test.sh/expected-output.txt for Python, JavaScript,
+      TypeScript (Bun), Perl, Racket, and C. All six verified locally.
+- [ ] **Step 2 — C++ and Go.** C++ needs GCC 14 (`<print>` header) — not on
+      the dev machine, so capture its output from the CI runner log on the PR,
+      then commit it as the expected file. Go is preinstalled on runners.
+- [ ] **Step 3 — Ruby, Lua, Haskell.** All installable on the runner via apt
+      (`ruby`, `lua5.4`, `ghc`); add installs to the workflow's toolchain step.
+- [ ] **Step 4 — OCaml, Elixir, Clojure, F#.** apt/preinstalled on runners
+      (`ocaml`, `elixir`, `clojure`; F# via preinstalled dotnet). Clojure runs
+      through its own deps.edn test runner — wrap it in test.sh.
+- [ ] **Step 5 — Java and Kotlin.** Java: PR #29 (the `java` branch) fixes the
+      numeral typing so PRED works; add test.sh + expected output to that
+      branch, let CI validate it, then merge #29. Kotlin: runner has a JDK;
+      install kotlinc in the workflow.
+- [ ] **Step 6 — exotic languages: ArkScript, FatScript, bruijn, Language 84.**
+      Each needs its own toolchain acquisition (GitHub releases, cargo/stack
+      installs, or building from source). If one is genuinely unobtainable in
+      CI, document that in its folder instead of leaving it silently untested.
+- [ ] **Step 7 — flip the default and update the front door.** Once all
+      folders have tests: make `run-tests.sh` fail on folders *missing* a
+      test.sh, update README contribution instructions to require
+      test.sh + expected-output.txt in new-language PRs, and enable branch
+      protection so PRs need a green check to merge.
+
+## Milestone 2 — fill out the missing major languages
+
+One language per step, each landing as its own PR with implementation, README,
+test.sh, and expected output, CI-green before merge. Order roughly by reach:
+
+- [ ] Rust
+- [ ] C#
+- [ ] Swift
+- [ ] Scala
+- [ ] Zig
+- [ ] PHP
+- [ ] Dart
+- [ ] Gleam (reopens the slot from closed PR #28)
+
+Keep a "wanted" list in the README for languages left open to contributors
+(Prolog, Erlang, APL, Idris, …).
+
+## Milestone 3 — housekeeping
+
+- [ ] Remove the committed binary `languages/c/c/app` and add a root
+      `.gitignore` for build artifacts.
+- [ ] TypeScript implementation only exercises Church numerals at runtime —
+      booleans exist at the type level only. Decide whether that satisfies the
+      spec or needs a runtime supplement.
+- [ ] Comment on merged PR #6 noting the `Function<Term, Term>` fix, closing
+      the loop on the thread there.
+- [ ] Java README still describes an Oracle JDK manual install — point it at
+      apt/Temurin and the test.sh instead.

languages/c/c/expected-output.txt

@@ -0,0 +1,2 @@
+1
+1

languages/c/c/test.sh

@@ -0,0 +1,5 @@
+gcc lambda-core.c -std=c89 -pedantic -o lambda-core.test || exit 1
+./lambda-core.test
+status=$?
+rm -f lambda-core.test
+exit $status

languages/j/javascript/expected-output.txt

@@ -0,0 +1,30 @@
+LOGIC
+---------------
+TRUE/FALSE
+t
+f
+NOT
+f
+t
+AND
+t
+f
+f
+f
+OR
+t
+t
+t
+f
+
+CHURCH NUMERALS
+---------------
+ZERO/SUCC
+0
+1
+2
+3
+PRED
+0
+1
+2

languages/j/javascript/test.sh

@@ -0,0 +1 @@
+node lambda-core.js

languages/p/perl/expected-output.txt

@@ -0,0 +1,24 @@
+LOGIC
+-------------
+ TRUE: true
+ FALSE: false
+ NOT TRUE: false
+ NOT FALSE: true
+ TRUE AND TRUE: true
+ TRUE AND FALSE: false
+ FALSE AND TRUE: false
+ FALSE AND FALSE: false
+ TRUE OR TRUE: true
+ TRUE OR FALSE: true
+ FALSE OR TRUE: true
+ FALSE OR FALSE: false
+
+CHURCH NUMERALS
+-------------
+ ZERO: 0
+ ONE: 1
+ SUCC ONE: 2
+ SUCC SUCC ONE: 3
+ PRED ONE: 0
+ PRED SUCC ONE: 1
+ PRED SUCC SUCC ONE: 2

languages/p/perl/test.sh

@@ -0,0 +1 @@
+perl lambda-core.pl

languages/p/python/expected-output.txt

@@ -0,0 +1,30 @@
+LOGIC
+-------------
+TRUE/FALSE
+t
+f
+NOT
+f
+t
+AND
+t
+f
+f
+f
+OR
+t
+t
+t
+f
+
+CHURCH NUMERALS
+-------------
+ZERO/SUCC
+0
+1
+2
+3
+PRED
+0
+1
+2

languages/p/python/test.sh

@@ -0,0 +1 @@
+python3 lambda-core.py

languages/r/racket/expected-output.txt

@@ -0,0 +1,30 @@
+LOGIC
+--------------
+TRUE/FALSE
+t
+f
+NOT
+f
+t
+AND
+t
+f
+f
+f
+OR
+t
+t
+t
+f
+
+CHURCH NUMERALS
+--------------
+ZERO/SUCC
+0
+1
+2
+3
+PRED
+0
+1
+2

languages/r/racket/test.sh

@@ -0,0 +1 @@
+racket lambda-core.rkt

languages/t/typescript/expected-output.txt

@@ -0,0 +1,8 @@
+Zero: 0
+One: 1
+Two: 2
+Three: 3
+Pred Zero: 0
+Pred One: 0
+Pred Two: 1
+Pred Three: 2

languages/t/typescript/test.sh

@@ -0,0 +1 @@
+bun run lambda-core.ts

run-tests.sh

@@ -0,0 +1,35 @@
+#!/bin/sh
+# Runs every language folder that has a test.sh and compares its stdout
+# against that folder's expected-output.txt. Exits nonzero if any differ.
+# Folders without a test.sh are skipped (not yet backfilled).
+set -u
+
+root=$(cd "$(dirname "$0")" && pwd)
+fail=0
+ran=0
+
+for script in "$root"/languages/*/*/test.sh; do
+    [ -e "$script" ] || continue
+    dir=$(dirname "$script")
+    name=${dir#"$root"/languages/}
+    ran=$((ran + 1))
+
+    actual=$( (cd "$dir" && sh test.sh 2>/dev/null) )
+    status=$?
+    expected=$(cat "$dir/expected-output.txt")
+
+    if [ "$status" -eq 0 ] && [ "$actual" = "$expected" ]; then
+        echo "PASS  $name"
+    else
+        echo "FAIL  $name (exit $status)"
+        echo "--- expected ---"
+        echo "$expected"
+        echo "--- actual ---"
+        echo "$actual"
+        echo "---"
+        fail=1
+    fi
+done
+
+echo "$ran folder(s) tested"
+exit $fail