* Add benchmark suite for alloc/dispatch/collection workloads (jolt-1r86)
The ray tracer is float-compute-bound (devirt, alloc removal, type-proving all
measured flat on it), so it can't validate the optimization passes. Add a small
cross-language suite (AWFY + CLBG style, portable Clojure) isolating the axes it
misses:
binary-trees allocation / GC pressure (escaping short-lived records)
dispatch megamorphic protocol dispatch (~1M dispatches/s; WP can't devirt)
collections persistent map/vector churn
bench/run.sh runs them; bench/README.md maps each to the pass it exercises.
collections immediately surfaced jolt-684u: the persistent hash map is O(n) per
assoc (flat copy-on-write bucket array, not a HAMT) — n=4000 assocs take 50s.
Invisible to the ray tracer (no maps).
* Persistent hash map: HAMT instead of O(n) copy-on-write (jolt-684u)
The map was a flat bucket array whose assoc copied the whole array every insert
(O(n)/assoc, O(n^2) to build). Compounding it, small maps are Janet structs that
only promoted to phm for collection keys — never for size — so a scalar-key map
stayed an O(n)-copy struct forever. Building a 4000-entry map took ~50s.
Two fixes, following ClojureScript's design:
- phm.janet is now a HAMT (hash array mapped trie): BitmapIndexedNode /
ArrayNode / HashCollisionNode, 32-way, 5 hash bits per level, structural
sharing — assoc/dissoc/get are O(log32 n). Translated from cljs.core, adapted
to Janet's 32-bit bit-ops (the hash is carried unsigned, the level index is
extracted with arithmetic, and bits are tested with band against 1<<i since
brushift rejects negative bitmaps). The public phm-* API and the value shape
(:jolt/type :jolt/phm, :cnt) are unchanged; transients are a separate rep and
untouched.
- core_coll promotes a struct map to a phm past 8 entries (not only for
collection keys), mirroring cljs PersistentArrayMap -> PersistentHashMap, so
incremental building isn't O(n^2).
20000 raw assocs: 7.1s -> 0.105s. The collections benchmark: 16.7s -> 0.2s.
Correctness covered by test/unit/phm-hamt-test.janet (oracle vs a Janet table,
nil keys, dissoc, a real hash-collision pair, and a sub-linear-assoc guard);
full gate green.
---------
Co-authored-by: Yogthos <yogthos@gmail.com>
5d: document the seed↔overlay boundary and add a drift check. core fns split
across a Janet seed (core-X, registered in core-bindings) and a Clojure overlay;
five names (char?/sorted?/sorted-map?/sorted-set?/transduce) carry a defn in
both, with the overlay copy authoritative and the seed copy internal-only. The
into-vs-transduce home asymmetry was undocumented. Adds docs/seed-overlay-registry.md,
SEED-TWIN: comments at the five seed sites, and a build-time drift check
(test/unit/seed-overlay-registry-test.janet) that recomputes the twin set from
source and fails if it diverges or a twin leaks into core-bindings.
5e: rep↔API pointer comments in pv/plist/phm/phs/lazyseq (representation lives
here; Clojure-facing ops dispatch in core_coll/core_types) and back-pointers in
core_coll. No behavior change — comments, docs, one source-analysis test.
Full gate green (suite ≥4695 pass / ≥88 clean files), drift check passes.
main.janet held ~45 lines of env-knob policy (open-mode / direct-link / optimize
/ shapes / whole-program gates) that couldn't be unit-tested without the CLI, and
two disk-image caches (api/init-cached, main/deps-image) each hand-built a
POSITIONAL "%q|%q|..." key that silently misaligned if a ctx-shaping knob was
added in only one place.
config.janet now owns:
ctx-shaping-env-vars the canonical list of env vars that shape the built ctx
ctx-cache-key a labeled key (name=value) over a prefix + every shaping
var, so adding a knob updates BOTH cache keys at once and
can't positionally alias two different builds
resolve-run-mode [open-mode? main-entry?] -> the ctx env knob map
main shrinks to: compute open-mode?/main-entry? from argv, call resolve-run-mode,
install the knobs. Both deps-image-path (main) and image-cache-path (api) build
their keys via ctx-cache-key. New test/unit/config-test.janet locks in the
run-mode cases and asserts every ctx-shaping env var participates in the key.
Scope: this is 4a + the 4b cache-key footgun fix. The optional 4b cleanup
(folding the load/save image dance + aot marshal helpers into one ctx_image
module) is left for a follow-up — it's lower value and higher blast radius.
No behavior change (cache keys now key on a superset of env vars, so at worst a
one-time cold rebuild). Gate green: conformance 355x3, clojure-test-suite 4718
pass (>= 4695 baseline), config-test, full jpm test exit 0.
* Protocol/interop fixes to run metosin/malli
Bringing up malli (schema validation) surfaced a batch of protocol and host-interop
gaps. m/validate now works across the schema vocabulary (predicates, :map incl.
nested/optional, :vector, :tuple, :enum, :maybe, :and, bounded int/string).
- extend-type and reify now accept MULTIPLE protocols in one form (each bare
symbol switches the current protocol). reify records every protocol it
implements, so instance?/satisfies? recognise all of them.
- Protocol method params support destructuring: reify/extend-type/deftype/
defrecord emit (fn ...) (which desugars patterns) instead of raw fn*.
- instance? of a PROTOCOL works like satisfies? for reify/record instances,
matching short names across qualified/bare protocol references.
- @x reads as the qualified clojure.core/deref, so it still derefs where a ns
excludes and rebinds deref (malli does). Updated reader-test + the reader
spec/grammar (S11, deref rule).
- Java collection interop on jolt collections: .nth/.count/.valAt/.get/.seq/
.containsKey route to the clojure.core equivalent (1-arg and 0-arg paths).
- java.util.HashMap capacity/load-factor constructors + .putAll.
- A class used as a value resolves to its instances' type, so Pattern -> the
regex type (malli keys class-schemas by it).
- Shims for malli's load path: LazilyPersistentVector/createOwning and
PersistentArrayMap/createWithCheck statics.
m/explain not yet working (jolt-fjb1). Full gate green.
* satisfies? recognizes reify, consistent with instance?
A reify's protocol methods are instance-local, so they aren't in the global type
registry that type-satisfies? consults — satisfies? returned false for a reify
even when it implemented the protocol. Check the protocols the reify records on
itself (the same :jolt/protocols list instance? uses), matching short names like
instance? does. Covers single- and multi-protocol reify.
---------
Co-authored-by: Yogthos <yogthos@gmail.com>
The phm had a FIXED 8 buckets, so a 100-entry map was a ~12-entry linear
scan per lookup — and phm-get walked the bucket twice (contains? then find).
This went mostly unnoticed until the canonical zipmap (batch 2) started
returning phms where kvs->map had built structs for scalar keys, regressing
the map-read bench ~7x (jolt-s3y).
phm-assoc now rehashes into a doubled bucket array when the count passes 2
entries/bucket (done on the fresh copy, so persistence is untouched);
phm-get is single-pass with a presence flag (nil values still distinguish
from missing); key= tries identity/scalar equality before paying for
canonicalization; the bucket count is derived from (length (m :buckets)),
not a constant, so any already-marshaled map keeps working. core-contains?'s
phm branch goes through phm-contains? instead of poking buckets directly.
map-read 48.5 -> 10.9 ms (the residual vs the pre-batch-2 6.7 is the
canonicalizing-representation constant); map-build steady; bench TOTAL 4457
vs 4565 on main back-to-back. New unit case crosses the resize boundary at
500 entries: every key found, nil values present, collection keys canonical,
dissoc + persistence intact. Gate: jpm exit 0, conformance 326x3.
init in compile mode is ~2.4 s (tier loading, analyzer self-compile, macro
recompilation), paid by every process that builds a ctx from source — each
jpm-test file, embedders, workers. init-cached marshals the built ctx to a
disk image (same root-env dicts as snapshot/fork) and later processes
unmarshal it in ~5 ms, any process: nothing from the baking process is
needed at load.
The cache key fingerprints the embedded .clj stdlib (which covers jolt-core:
analyzer, IR, core tiers), the .janet seed sources next to the module, the
janet version, the init opts, and the env knobs that shape a ctx (JOLT_PATH/
MUTABLE/AOT_CORE/FEATURES) — any change rebuilds. Corrupt or non-ctx images
fall back to a rebuild (unmarshal of garbage can 'succeed' with a scalar, so
the shape is checked, not just the throw). Writes are atomic (tmp + rename)
so racing cold starts never publish a torn image. JOLT_NO_IMAGE_CACHE=1
opts out; JOLT_IMAGE_CACHE_DIR overrides the location (default TMPDIR).
Test consumers switch to init-cached (harness, suite-worker, conformance,
the behavioral unit/integration tests); tests that validate the bootstrap
itself (bootstrap-fixpoint, staged-bootstrap, aot round-trip, direct-linking)
and the deps tests (tmp-dir :paths would fragment the key) keep real init.
Full jpm test: 2:46 -> 1:58 (~29%). New ctx-image-test covers cold/warm,
cross-process load (subprocess runs defn/redef/macros/protocols/multimethods
off the baked image), per-opts keying, and corrupt-image fallback.
The 1104-line Janet bootstrap compiler existed to build jolt.ir/jolt.analyzer
and the kernel tier before the self-hosted analyzer could exist. It is
replaced by the interpreter + one fixpoint turn:
1. bootstrap-load-source loads the compiler sources INTERPRETED (the
evaluator can run the analyzer — it always could).
2. After the overlay is up, self-compile-compiler! re-runs the kernel tier,
jolt.ir, and jolt.analyzer through the SELF-HOSTED pipeline — the
interpreted analyzer compiles itself, and steady state runs compiled with
no bootstrap compiler involved.
Measured: init {:compile? true} 1093 -> ~2400 ms (the one-time interpreted
pass + self-compilation), but steady-state compilation is 2.8x FASTER
(100 forms: 134 -> 48 ms) — the self-hosted pipeline emits better code than
the bootstrap did. An AOT image for init cost is future work (aot.janet's
machinery is the natural vehicle).
The bootstrap's runtime kernel moves to backend.janet (jolt-runtime-env,
ctx-janet-env, build-map-literal); aot imports it from there. The
uncompilable-error? punt check unwraps the interpreter's exception struct
(the interpreted analyzer's throw arrives wrapped). compile-string/
compile-file (the bootstrap's source-text emitter API, no callers outside
the bootstrap's own unit test) are removed with it, as is compiler-test.
Gate green across everything incl. fixpoint stage1==2==3, AOT round-trip,
uberscript, CLI; conformance 326x3; suite 4566 >= 4540; bench in band.
resolve-sym's last resort silently resolved any unknown Clojure symbol
against Janet's root environment — leaking Janet builtins with JANET
semantics into Clojure code: (type 1) was Janet's :number, (gensym) returned
Janet symbols (the long-documented (symbol (str (gensym))) macro landmine
existed BECAUSE of this), compare/slurp/int?/any? likewise. The explicit
janet/ prefix is the deliberate interop channel; the implicit fallback is
gone — an unresolved symbol is an error.
What the leak was masking, now proper interned vars:
- gensym: jolt's own (already existed, never interned) — returns real jolt
symbols; the macro landmine is dead
- compare: full Clojure total order (nil-first, numbers, strings, keywords,
symbols by ns/name, booleans, chars, uuid/inst, vectors by length then
elementwise; cross-type throws)
- type: :type metadata override, deftype/record tag as symbol, else a
taxonomy keyword (host-classified)
- int?: core-integer? — which had a latent bug the leak hid: (integer?
##Inf) was true (floor of inf is inf); NaN/infinities now excluded
- any?: constantly true (Clojure 1.9; SCI's namespaces.cljc needs it)
- jolt.interop/janet-type now uses the explicit (janet/type x) channel
- evaluator-test uses init (a bare make-ctx resolved EVERYTHING via the leak)
Suite 4470 -> 4532+ pass / 86-87 clean (proper compare unlocks the sort
files); baselines raised. Conformance 326x3 (+5 rows), +22 predicate spec
rows, stdlib battery green, all specs+unit. Coverage dashboard now counts
previously-leak-resolvable names honestly (missing-portable 19 -> 27).
jolt no longer satisfies :clj in reader conditionals. The shortcut was a
measured net liability: :clj branches carry JVM interop and JVM-specific
test expectations jolt fails, and they shadowed :default branches jolt
passes. A/B over the suite: clj,default = 4967 assertions / 4324 pass / 119
errors; jolt,default = 5069 / 4470 / 81 (+146 pass, -38 errors, +8 clean
files). Baselines raised to 4470/86.
Matching is now by CLAUSE order like Clojure — the first clause whose key is
in the feature set wins (#?(:default 5 :clj 6) is 5 everywhere); the old code
scanned for :clj first, then :default, regardless of position.
Foreign clj-targeted libraries are a property of the LOADING CONTEXT, not the
platform: reader-features-set! opts a load into a compatibility set, and the
SCI bootstrap/runtime tests load SCI under ["jolt" "clj" "default"] (its
.cljc selects implementations via :clj with no :jolt branches).
JOLT_FEATURES remains the process-wide override.
RFC 0002 records the decision with the measured data; spec 02-reader S18 is
now normative (clause order, documented feature set, per-context override).
Reader tests updated to the portable set + an opt-in round-trip.
prefer-method/remove-method/remove-all-methods/get-method/methods become
overlay macros over ctx-capturing *-setup fns (a multimethod's method table
lives on its VAR, so the name passes quoted — the defmulti/defmethod shape).
instance? likewise (class names don't evaluate to values); satisfies? is a
plain ctx-capturing fn (evaluated args). locking and defonce become overlay
macros — locking now also evaluates the monitor expr (the old arm skipped it
and any body form past the first); defonce keeps the existing-root check.
read-string and macroexpand-1 are ctx-capturing fns.
Removed all eleven from the evaluator special arms + special-symbol?,
host_iface special-names, and compiler uncompilable-heads. evaluator-test's
locking/instance? cases use init now (overlay macros need the full env).
Surfaced pre-existing (filed): multimethod dispatch records prefer-method
preferences on the var but never consults them in ambiguous isa dispatch.
Gate: conformance 296x3 (+11 tier-6c cases), fallback-zero 73/3, fixpoint,
self-host, sci, staged, suite 4049>=4034, all specs+unit.
compile-and-eval wrapped the compile step in a blanket protect — any failure,
including a genuine compiler bug, silently fell back to the interpreter and
hid behind a correct-looking result. Now only the analyzer's deliberate punt
signal (jolt/uncompilable: …) falls back; any other compile-step error
propagates.
Tightening this surfaced one accidental dependency: pre-kernel overlay forms
(00-syntax's destructure defn) trigger a compile while ensure-analyzer is
still gated, and the missing analyzer crashed var-get with a cryptic indexing
error that the blanket catch happened to convert into the designed interpret
fallback. That path now punts explicitly ("jolt/uncompilable: analyzer not
built (pre-kernel bootstrap)").
New unit test stubs jolt.analyzer/analyze to prove a non-punt error
propagates while the punt channel still falls back.
Gate: conformance 269x3, fallback-zero, fixpoint stage1==2==3, self-host,
sci-bootstrap, staged-bootstrap, all specs+unit, suite 4046>=4034 (5 timeouts);
core-bench A/B neutral (4562 vs 4572 ms).
* core: Stage 2 Task 2 tier 5a — compile defmulti + defmethod
defmulti/defmethod become macros (30-macros) over ctx-capturing
clojure.core fns (defmulti-setup/defmethod-setup, interned by
install-stateful-fns!):
- defmulti: (defmulti name dispatch & opts) -> (defmulti-setup 'name
dispatch ~@opts). name quoted; dispatch + opts (:default/:hierarchy)
evaluated. defmulti-setup builds the dispatch closure over the method
table and interns the var (same hierarchy/default/cache behavior).
- defmethod: (defmethod mm dval & fn-tail) -> (defmethod-setup 'mm dval
(fn ~@fn-tail)). The method impl is now a COMPILED (fn …) (was an
interpreted fn* eval). Auto-creates the multimethod if missing.
- removed their special-symbol? entries + eval-list arms, and dropped them
from host_iface special-names + loader stateful-head?.
Both compile + interpret as plain invokes; dispatch incl. :default and
derive/hierarchy works in both modes.
Tests: evaluator-test (defmulti case) + namespace-test now use init (these
forms are overlay macros now, so a bare make-ctx lacks them).
Gate green: conformance 269x3, fallback-zero 38/4, bootstrap-fixpoint
stage1==2==3, self-host, staged-bootstrap, sci-bootstrap, clojure-test-suite
>=4034/67, features 78/78, all unit + spec (multimethods 16/16).
* core: Stage 2 Task 2 tier 5b — compile deftype + defrecord
deftype becomes a macro (30-macros) over make-deftype-ctor (a ctx-capturing
clojure.core fn that bakes the ns-qualified type tag at def time) plus
extend-type for any inline protocol methods — so it compiles as a plain (do …).
Mirrors defrecord's existing field-let/protocol-grouping pattern.
- make-deftype-ctor-impl (evaluator) builds the ctor; interned as a closure.
- removed the deftype special-symbol? entry + eval-list arm; dropped deftype/
defrecord from host_iface special-names + loader stateful-head?.
- defrecord no longer redefines ->name via (Name. …) interop (frozen) — deftype
already provides ->name, so defrecord compiles too (map->name builds via it).
- field-kws spliced into a vector LITERAL ([~@…]) so the analyzer sees a vector
form, not a runtime pvec; type name + fields are unwrapped of ^meta (the reader
yields (with-meta sym m) forms, e.g. sci's (deftype ^{:doc …} Var …)).
With tier 5a, all of deftype/defrecord/defmulti/defmethod compile. The loader's
interpret-only set is now just the frozen host-coupled forms: defmacro/set!/./
new/eval.
Tests: evaluator-test deftype case uses init (deftype is an overlay macro now);
fallback-zero moves deftype off must-punt, adds deftype/defrecord/defmulti/
defmethod to must-compile (43/3).
Gate green (full jpm build + jpm test): conformance 269x3, fallback-zero 43/3,
bootstrap-fixpoint stage1==2==3, self-host, staged-bootstrap, sci-bootstrap
422/0, clojure-test-suite >=4034/67, all unit + spec.
---------
Co-authored-by: Yogthos <yogthos@gmail.com>
adds self-hosted compiler is functionally:
- The default compile path is the portable pipeline using jolt.analyzer (Clojure) → host-neutral IR → backend.janet.
- The analyzer is itself Clojure, compiled by jolt for true self-hosting.
- bootstrap-fixpoint passes (stage1 == stage2 == stage3): rebuilding the compiler on its own output.
- clojure.core is now self-hosted in the overlay.
- Stateful forms (defmacro/ns/deftype/defmulti/require/in-ns) are interpreted by design.
The reader expanded ^X form to (with-meta form X), which evaluated the tag (so
^String errored 'Unable to resolve symbol: String') and, as a param, was no
longer a bare symbol (so the arg bound to nil). Now a keyword/symbol/string hint
on a symbol attaches to the symbol's :meta and the symbol stays bare, so type
hints are transparent in params, lets, and bodies. Map metadata still uses a
runtime with-meta form.
meta now returns a symbol's :meta, and def applies the name's metadata
(^:dynamic, ^:private, ^Type tag, ^{:doc}) to the var, so (meta (var x)) is
consistent. Specs in metadata-spec; grammar note in the ebnf.
README notes regex \p{...} as unsupported (separate from this).
Two changes unlock native Janet speed in compile mode:
- Hot numeric primitives (+ - * < > <= >=) emit as native Janet SYMBOLS rather
than the variadic core fns, so Janet's compiler uses its arithmetic/compare
opcodes. = / not= / quot / rem / mod / division stay as core fns (their
semantics differ from Janet's). Trade-off: the strict non-number checks are
relaxed under compilation (documented perf-mode divergence).
- emit-invoke emits a DIRECT call (f arg...) when the callee is a function
reference (core/local/symbol/fn), instead of wrapping every call in jolt-call.
jolt-call is kept only for keyword/collection literals in call position
((:k m), ({:a 1} :a)) so IFn dispatch still works.
compiled fib(30): 3.4s -> 0.076s (native ceiling), faster than jank's 0.8s.
Updated compiler-test string assertions (core-+ -> +); compile-mode-test gains
native-op + IFn-dispatch cases; README documents compile mode. jpm test green.
Reorganize the flat 49-file test/ into three layers (jpm test recurses, so all
are still discovered):
- test/unit/ white-box component tests (reader, evaluator, types,
persistent-map, lazy-seq, macro, interop, compiler)
- test/integration/ cross-cutting + regression batteries (conformance, jank,
sci-bootstrap/runtime, features, systematic-coverage, api,
core, namespaces, ported clojure suites) and
.../ports/ ported clojure/cljs test batches pending consolidation
- test/spec/ the behavioral contract (built out in following commits)
- test/support/harness.janet shared defspec table runner (cases compared via
Jolt's own =, with a :throws sentinel) + expect= helpers
Files moved with git mv (history preserved) and import paths fixed for depth.
jpm test green. README Test section updated.
Next: build out test/spec/ to cover the public API area-by-area, mining the
integration batteries and filling gaps.
