dsh-vibe-math 2.2.2 → 2.3.1

package/run-tests.mjs

@@ -0,0 +1,99 @@
+#!/usr/bin/env node
+/**
+ * PARALLEL TEST RUNNER — run every shipped suite (or a filtered subset) concurrently and report
+ * per-suite timings, so the strategy for the next run is chosen from DATA instead of guesswork.
+ *
+ * Why this exists: the suites are wildly uneven (one suite is ~160 s, most are under 2 s), so a
+ * sequential sweep spends almost all of its wall time waiting for the slowest one. Running them
+ * with a worker pool makes the sweep bounded by the slowest SUITE rather than by their SUM.
+ * On this machine (4 cores) the sweep went from ~5.5 min to ~2 min; see docs/test-timing.md.
+ *
+ * Every suite already isolates itself (each creates its own mkdtemp workspace), so parallelism is
+ * safe. Suites that WRITE a corpus take a per-run corpus dir from an env var; this runner points
+ * them at a scratch dir when it runs a suite more than once, which it never does — but the probe
+ * runner (audit-formal-sensitivity.mjs) does, and it passes its own dirs.
+ *
+ * Usage:
+ *   node run-tests.mjs                      # every *.test.mjs, concurrency = min(4, cpus)
+ *   node run-tests.mjs --concurrency=6
+ *   node run-tests.mjs --only formal        # substring match on the file name (repeatable, OR)
+ *   node run-tests.mjs --exclude e2e-v4     # substring to skip (repeatable)
+ *   node run-tests.mjs --json               # machine-readable summary on stdout
+ */
+import { spawn } from 'node:child_process'
+import { readdirSync } from 'node:fs'
+import { cpus } from 'node:os'
+import { join } from 'node:path'
+import { fileURLToPath } from 'node:url'
+
+const HERE = fileURLToPath(new URL('./', import.meta.url))
+const argv = process.argv.slice(2)
+const flag = (name) => argv.filter((a) => a.startsWith('--' + name + '=')).map((a) => a.split('=').slice(1).join('='))
+const has = (name) => argv.includes('--' + name)
+const only = flag('only')
+const exclude = flag('exclude')
+const asJson = has('json')
+const concurrency = Math.max(1, Number(flag('concurrency')[0] || Math.min(4, cpus().length)))
+
+let suites = readdirSync(HERE).filter((f) => f.endsWith('.test.mjs')).sort()
+if (only.length) suites = suites.filter((f) => only.some((o) => f.includes(o)))
+if (exclude.length) suites = suites.filter((f) => !exclude.some((o) => f.includes(o)))
+if (!suites.length) { console.error('no suites matched'); process.exit(2) }
+
+function runSuite(file) {
+  return new Promise((resolve) => {
+    const t0 = Date.now()
+    const child = spawn(process.execPath, [file], { cwd: HERE, stdio: ['ignore', 'pipe', 'pipe'] })
+    let out = '', err = ''
+    child.stdout.on('data', (d) => { out += d.toString() })
+    child.stderr.on('data', (d) => { err += d.toString() })
+    child.on('error', (e) => resolve({ file, code: -1, ms: Date.now() - t0, out, err: err + '\n' + String(e) }))
+    child.on('close', (code) => resolve({ file, code, ms: Date.now() - t0, out, err }))
+  })
+}
+
+const results = []
+let cursor = 0
+const started = Date.now()
+async function worker(id) {
+  for (;;) {
+    const i = cursor++
+    if (i >= suites.length) return
+    const r = await runSuite(suites[i])
+    const tail = String(r.out).trim().split('\n').filter(Boolean).slice(-1)[0] || ''
+    results[i] = r
+    const mark = r.code === 0 ? 'PASS' : 'FAIL'
+    console.log(
+      mark + '  ' + r.file.padEnd(38) +
+      ' exit=' + String(r.code).padStart(3) +
+      '  ' + (r.ms / 1000).toFixed(1).padStart(6) + 's' +
+      (tail ? '  ' + tail.slice(0, 78) : '')
+    )
+    if (r.code !== 0) {
+      const lines = (r.out + '\n' + r.err).split('\n').filter(Boolean)
+      for (const l of lines.slice(-15)) console.log('      ' + l)
+    }
+  }
+}
+await Promise.all(Array.from({ length: Math.min(concurrency, suites.length) }, (_, i) => worker(i)))
+
+const wall = (Date.now() - started) / 1000
+const sum = results.reduce((a, r) => a + r.ms, 0) / 1000
+const bad = results.filter((r) => r.code !== 0)
+const slowest = results.slice().sort((a, b) => b.ms - a.ms).slice(0, 5)
+
+if (asJson) {
+  console.log(JSON.stringify({
+    concurrency, wallSeconds: Number(wall.toFixed(1)), sumSeconds: Number(sum.toFixed(1)),
+    pass: results.length - bad.length, fail: bad.length,
+    suites: results.map((r) => ({ file: r.file, exit: r.code, seconds: Number((r.ms / 1000).toFixed(1)) })),
+  }, null, 2))
+} else {
+  console.log('')
+  console.log('concurrency ' + concurrency + '  ·  wall ' + wall.toFixed(1) + 's  ·  sum of suite times ' + sum.toFixed(1) + 's'
+    + '  ·  speed-up x' + (sum / Math.max(wall, 0.001)).toFixed(2))
+  console.log('slowest: ' + slowest.map((r) => r.file.replace('.test.mjs', '') + ' ' + (r.ms / 1000).toFixed(1) + 's').join('  ·  '))
+  console.log('TOTAL ' + results.length + '  PASS ' + (results.length - bad.length) + '  FAIL ' + bad.length)
+  for (const b of bad) console.log('  FAILED: ' + b.file + ' (exit ' + b.code + ')')
+}
+process.exit(bad.length === 0 ? 0 : 1)

package/vibe-math-v2/agent.cordis.yml

@@ -28,11 +28,14 @@
       - vibe_math_status / vibe_math_report — read scheduler status / full progress report (report also writes Progress_Logs/report.json).
       - vibe_math_pause / vibe_math_abort — pause / abort (abort interrupts all children).
       - vibe_math_set_mode {mode: manual|auto} — switch manual / auto control.
-      - vibe_math_set_params {...} — tune any parameter (see vibe_math_setup for the full schema; e.g. reportMode file|push|both, promoteValueThreshold, verdictMode flat|forced).
+      - vibe_math_set_params {...} — tune any parameter (see vibe_math_setup for the full schema; e.g. reportMode file|push|both, promoteValueThreshold, verdictMode flat|forced, formalVerify off|encourage|require).
       - vibe_math_setup / vibe_math_save_settings / vibe_math_template — guided configuration / persist defaults / generate template.
       - vibe_math_new_project / vibe_math_set_project / vibe_math_list_projects — per-project folders.
       - vibe_math_list_decisions / vibe_math_decide {id, action: approve|reject|override, verdict?} — resolve manual decisions.
       - vibe_math_list_agents / vibe_math_message_agent / vibe_math_interrupt_agent — inspect / steer / interrupt subagents.
+      - vibe_math_lean_run / vibe_math_lean_archive / vibe_math_lean_lib — Lean formal
+        verification (execute / archive / list the reuse library). The scheduler's child agents
+        use them too; they work in every mode.
 
       A /vibe slash command mirrors the main controls. Data lives under {{cwd}}/VibeMath/Projects/<project>/
       (qs/qs.json, Propos/<分类>_Propos.json, Reliable/, Verified/, Verification_logs/, Progress_Logs/, VibeMath_State/)
@@ -53,6 +56,22 @@
       with its proofs/refutations transferred into the solution list (verification results sync back to the
       source proposition); a solver-reported sub-question q_sub registers THREE objects — the q_sub problem,
       the temporary-assumption proposition p_{q-tmp}, and the problem "判断下述命题是否成立：p_{q-tmp}".
+
+      LEAN FORMAL VERIFICATION (formalVerify, a tunable parameter):
+        - 'off' (default, no extra requirement) | 'encourage' (solver/verifier agents decide by
+          implementation difficulty whether to formalize in Lean; once a Lean run passes, the review
+          subject becomes FIDELITY — do the Lean definitions/objects/conditions/assumptions/conclusion
+          match the proposition as stated) | 'require' (same, plus a gate: a true/false verdict is
+          recorded as 未定论 with reason formal-required until the object is Lean-passed or carries an
+          explicit, reasoned blocker record; the scheduler is never wedged by it).
+        - Paths: work file Formal/<id>.lean; archived proof Verified/Lean/<id>.lean; reusable
+          definitions VibeMath/Formal/Lib/; proved lemmas VibeMath/Formal/Proved/.
+        - The toolchain knobs leanCommand / leanArgs / leanTimeoutMs are tunable as well
+          (e.g. leanCommand='lake' with leanArgs=['env','lean']); a missing Lean binary is
+          reported as LEAN_NOT_FOUND and still lets the code be written and archived.
+        - vibe_math_status / vibe_math_report show the mode, the per-object formal status and the
+          formalization TODO (Formal/TODO.md). The framework never installs Lean and never judges
+          fidelity for you.
     suffix: Your working directory is {{cwd}}.
     text: |-
       You are a coding agent powered by the {{model}} model. Your working directory is {{cwd}}.
@@ -72,11 +91,14 @@
       - vibe_math_status / vibe_math_report — read scheduler status / full progress report (report also writes Progress_Logs/report.json).
       - vibe_math_pause / vibe_math_abort — pause / abort (abort interrupts all children).
       - vibe_math_set_mode {mode: manual|auto} — switch manual / auto control.
-      - vibe_math_set_params {...} — tune any parameter (see vibe_math_setup for the full schema; e.g. reportMode file|push|both, promoteValueThreshold, verdictMode flat|forced).
+      - vibe_math_set_params {...} — tune any parameter (see vibe_math_setup for the full schema; e.g. reportMode file|push|both, promoteValueThreshold, verdictMode flat|forced, formalVerify off|encourage|require).
       - vibe_math_setup / vibe_math_save_settings / vibe_math_template — guided configuration / persist defaults / generate template.
       - vibe_math_new_project / vibe_math_set_project / vibe_math_list_projects — per-project folders.
       - vibe_math_list_decisions / vibe_math_decide {id, action: approve|reject|override, verdict?} — resolve manual decisions.
       - vibe_math_list_agents / vibe_math_message_agent / vibe_math_interrupt_agent — inspect / steer / interrupt subagents.
+      - vibe_math_lean_run / vibe_math_lean_archive / vibe_math_lean_lib — Lean formal
+        verification (execute / archive / list the reuse library). The scheduler's child agents
+        use them too; they work in every mode.
 
       A /vibe slash command mirrors the main controls. Data lives under {{cwd}}/VibeMath/Projects/<project>/
       (qs/qs.json, Propos/<分类>_Propos.json, Reliable/, Verified/, Verification_logs/, Progress_Logs/, VibeMath_State/)
@@ -98,6 +120,22 @@
       source proposition); a solver-reported sub-question q_sub registers THREE objects — the q_sub problem,
       the temporary-assumption proposition p_{q-tmp}, and the problem "判断下述命题是否成立：p_{q-tmp}".
 
+      LEAN FORMAL VERIFICATION (formalVerify, a tunable parameter):
+        - 'off' (default, no extra requirement) | 'encourage' (solver/verifier agents decide by
+          implementation difficulty whether to formalize in Lean; once a Lean run passes, the review
+          subject becomes FIDELITY — do the Lean definitions/objects/conditions/assumptions/conclusion
+          match the proposition as stated) | 'require' (same, plus a gate: a true/false verdict is
+          recorded as 未定论 with reason formal-required until the object is Lean-passed or carries an
+          explicit, reasoned blocker record; the scheduler is never wedged by it).
+        - Paths: work file Formal/<id>.lean; archived proof Verified/Lean/<id>.lean; reusable
+          definitions VibeMath/Formal/Lib/; proved lemmas VibeMath/Formal/Proved/.
+        - The toolchain knobs leanCommand / leanArgs / leanTimeoutMs are tunable as well
+          (e.g. leanCommand='lake' with leanArgs=['env','lean']); a missing Lean binary is
+          reported as LEAN_NOT_FOUND and still lets the code be written and archived.
+        - vibe_math_status / vibe_math_report show the mode, the per-object formal status and the
+          formalization TODO (Formal/TODO.md). The framework never installs Lean and never judges
+          fidelity for you.
+
 - id: agent-instructions
   name: '@deepseek-ai/dsh-agent-instructions'
   config: