jolt/test/chez/README.md
Yogthos edfd67a322 Vendor clojure-test-suite as a standing gate (make cts)
jank-lang/clojure-test-suite (per-core-fn clojure.test suites shared across
Clojure dialects) joins the default gate as vendor/clojure-test-suite, run by
host/chez/cts.sh: one joltc process per test namespace (a hang or crash is
contained by a per-process timeout), through the test/chez/cts-app project and
its cts-run runner, parallel workers.

Gating is exact per namespace against test/chez/cts-known-failures.txt, like
certify's allowlist: a namespace doing worse than the baseline fails, and one
doing better also fails as stale until the baseline is updated in the same
change. JOLT_CTS_WRITE_BASELINE=1 regenerates it; JOLT_CTS_NS runs a subset
verbosely.

Current standing: 243 namespaces, 5302 assertions pass, 340 fail + 236 error
across 88 namespaces pinned in the baseline (dominant clusters: BigDecimal
arithmetic operands, derive/ancestors hierarchy, transients, special-symbol?,
clojure.string case fns, the accepted narrow-int and seq-type-model
divergences). Two consecutive full runs produce identical counts. Wired into
make ci; skips cleanly when the submodule isn't checked out.
2026-07-02 03:46:57 -04:00

51 lines
2.7 KiB
Markdown

# Chez test harness
The correctness gate for jolt. Pure Chez (+ Clojure for the JVM oracle).
Correctness is judged against the JVM-sourced conformance spec; the spec itself
lives in `test/conformance/` (see its `SPEC.md`). Run the whole gate with `make
test` from the repo root.
## The spec corpus
`corpus.edn` is the contract: ~2920 rows `{:suite :label :expected :actual}`, with
`:expected` sourced from reference JVM Clojure by `test/conformance/regen-corpus.clj`.
It is frozen (the canonical source) — add or change cases here, then re-source the
answers with `regen-corpus.clj` and re-certify with `test/conformance/certify.clj`.
## The gate runners (`host/chez/`)
- `run-corpus.ss` — runs every corpus case through the spine (read → analyze → IR →
emit → eval, all on Chez), comparing each result by value-equality against the JVM
`:expected`. A `known-fail` allowlist covers cases jolt can't match because Chez has
no JVM host (Java classes, arrays, `BigDecimal`, opaque host-object printers, …);
the gate fails only on a NEW divergence or if the pass count drops below the floor.
chez --script host/chez/run-corpus.ss
JOLT_CORPUS_LIMIT=200 … # every-Nth stride, fast iteration
JOLT_CHEZ_ZJ_FLOOR=N … # override the floor (default 2678)
- `run-unit.ss` — host-specific unit cases (`test/chez/unit.edn`) that aren't in the
JVM-portable corpus: dot-forms, java statics, io, reader, walk, vars/namespaces,
refs. Each `:expr` is evaluated in-process and its printed value compared to a baked
`:expected` (`:throws` asserts a raise).
- `selfcheck.sh` — self-host fixpoint: `bootstrap.ss` rebuild byte-equals the
checked-in seed (`host/chez/seed/`).
- `smoke.sh` — real `bin/joltc -e` CLI smoke.
- `cts.sh` — the vendored [jank-lang/clojure-test-suite](https://github.com/jank-lang/clojure-test-suite)
(`vendor/clojure-test-suite`, a per-core-fn clojure.test suite shared across
Clojure dialects), run one namespace per `joltc` process (a hang or crash is
contained) through the `test/chez/cts-app` project and `cts-run` runner.
Per-namespace fail/error counts must exactly match the checked-in baseline
`test/chez/cts-known-failures.txt` — a namespace doing worse fails the gate,
and one doing better fails as stale until the baseline is updated in the same
change. `make cts`;
`JOLT_CTS_NS=ns1,ns2` runs a subset verbosely,
`JOLT_CTS_WRITE_BASELINE=1` regenerates the baseline.
## Other Chez tests
- `values-test.ss` — the value model (nil/truthiness/collections). `make values`.
- `bench-chez.ss` — compute bench through the pipeline (opt-in; not in the gate).
All runners assume `chez` on PATH.