Go has the same structure as erlang: suites load into one session and each
exposes a pass counter plus a *count* (total) counter rather than a fail
counter. MODE=dict fits — each suite's runner is a dict literal
{:passed P :failed (- count P) :total count}. No driver change; conformance.conf
+ 3-line shim, historical scoreboard schema preserved.
Parity verified 609/609 (0 fail), every suite matching baseline.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
forth's conformance.sh reads a foreign Forth test corpus (Hayes Core core.fr),
preprocesses it with awk + an external python3 chunk-splitter that generates a
chunks.sx of raw source strings, then runs them through the interpreter via
(hayes-run-all). The shared driver only epoch-loads SX preloads and evals SX
test suites emitting a counter/dict scoreboard — it cannot reproduce the
external preprocessing pipeline over a foreign .fr corpus (same category as
lua/smalltalk). No SX tests/*.sx suites exist to migrate. Left conformance.sh
untouched; recorded the exclusion.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Feed is the canonical MODE=counters shape: each suite runs in a fresh session
with shared preloads and a single feed-test-pass/feed-test-fail pair. Lifted the
old script's inline epoch-2 counter + feed-test helper defs into
lib/feed/test-harness.sx (preloaded last) so the driver can load them before
each suite. conformance.conf + 3-line shim; historical scoreboard schema
preserved. No driver change needed.
Parity verified 189/189 (0 fail), every suite matching baseline.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Erlang's suites load into one session and each exposes a pass counter plus a
*count* (total) counter rather than a fail counter, so MODE=dict fits directly:
each suite's runner is a dict literal {:passed P :failed (- count P) :total count}.
No driver change needed (dict mode already supports arbitrary runner expressions).
conformance.conf + 3-line shim; historical scoreboard schema preserved.
Parity verified 761/761 (0 fail), every suite matching baseline.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Extend the shared driver's MODE=counters with a backward-compatible SUITES
format: name:file[:pass-var:fail-var[:extra-preload ...]]. Optional per-suite
counter symbols (override the global COUNTERS_PASS/COUNTERS_FAIL) and per-suite
preload chains (loaded after the global PRELOADS). Plain name:file entries are
unchanged — verified against haskell (fib/sieve/quicksort 2/2/5, matches
committed scoreboard).
common-lisp has 8 distinct per-suite counter pairs and a different preload
chain per suite, so it could not fit the single-counter/fixed-preload model;
the extended format expresses it directly. conformance.conf keeps the historical
scoreboard schema; conformance.sh becomes the 3-line shim.
Result 487/487 (0 fail) vs the old 305/0 baseline — higher and explained: the
old per-suite 'timeout 30' was too tight for the slow eval suite (~15-25s under
contention), silently recording it as 0; the driver's 180s budget recovers its
true 182. geometry/mop-trace stay 0/0 (pre-existing refl-class-chain-depth-with
load error; counter vars defined as 0 -> clean gc-result, no fail-fallback).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Classified migratable-in-kind (SX suites over epoch, not a foreign runner)
but blocked on driver feature gaps: 8 distinct per-suite counter variable
name pairs and per-suite preload chains, neither supported by MODE=counters
(single global counter + fixed preloads) nor MODE=dict (load-time counter
collisions across suites). Baseline 305/0 across 12 suites. Did not migrate;
conformance.sh left untouched. Driver unchanged (out of per-iteration scope).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Reading lib/mod (Prolog) and lib/acl (Datalog) side by side shows the convergence
is in module names only. Federation: opposite trust models (SX registry + decision
sharing vs in-engine Datalog trust facts + fact replication), zero shared code.
Audit: only a ~5-fn core overlaps and it diverges (entry shapes, seq base 0 vs 1,
op sets, mutation idiom) — not worth a shared module under two restricted envs.
Outcome: keep them parallel; revisit only on a third same-model consumer.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
