The names turn 2a's leak removal exposed as honestly missing, now proper:
- slurp/spit/flush (host-classified): path-based IO over Janet files; spit
takes :append; flush flushes *out*. printf prints formatted (no newline)
over the existing format. file-seq walks paths via two host dir primitives
through the overlay's tree-seq.
- ns-map / ns-unmap / ns-refers (ctx fns). ns-refers required fixing the
refer MODEL: refer/use/:refer now map the SOURCE VAR into the target ns
(the Clojure model) instead of copying its value into a new var — so
source-ns redefinitions propagate, the :macro flag travels for free, and
refers are identifiable by the var's home :ns.
- Thread-binding family: with-bindings*/with-bindings, bound-fn*/bound-fn,
bound?, thread-bound?, get-thread-bindings. The captured binding map is a
Janet struct keyed by the var tables — the exact frame representation
var-get reads — so it re-pushes correctly (a phm frame is invisible to
var lookup).
- load-string and eval interned as VALUES at the api layer (they need the
loader's compile-or-interpret routing); the eval special form still
handles direct calls.
Suite 4532 -> 4572 (baseline floor 4540 across timeout variance, clean 86),
conformance 326x3, stdlib battery, all specs+unit (+21 turn-2b rows).
Coverage: missing-portable 27 -> 10 (left: the *in*-model readers, the
with-local-vars/with-precision/extenders tail).
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).
#inst (jolt-rnh): an instant is an immutable tagged struct
{:jolt/type :jolt/inst :ms <epoch-millis>} — equality and map-key hashing by
INSTANT, so different offsets denoting the same moment are =. The reader
parses RFC3339 with Clojure's partial-timestamp defaults (#inst "2020" is
2020-01-01T00:00:00.000Z) and errors on malformed input; inst?/inst-ms in
the overlay; pr-str prints the canonical
#inst "yyyy-MM-ddThh:mm:ss.fff-00:00" and round-trips; str gives the bare
RFC3339 string. Self-evaluating in the evaluator (like uuid/chars).
Syntax-quote (jolt-l2a): a syntax-quoted self-evaluating literal (string,
number, boolean, nil, keyword, char) collapses to the literal at READ time,
matching Clojure's reader — so nested/adjacent backticks over literals are
inert: (= "meow" ```"meow") is true (jank pass-adjacent). Symbols still
qualify and collections still template. General nested syntax-quote over
non-literals remains UNVERIFIED in spec S25.
Tests: inst-spec (18 cases incl. partial defaults, offsets, round-trip),
+10 literal-collapse reader rows, +5 conformance rows (321x3). Spec
02-reader S20/S25 updated to normative. Suite stable 4470/86.
One documentation root: the prose docs (building-and-deps, self-hosting
architecture/compiler, tools-deps, grammar.ebnf) join the spec and RFCs under
docs/. References in README and deps-conformance-test updated.
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.
The lexical-syntax chapter, granularity modeled on jank's 62-file
per-construct reader corpus: token grammar (whitespace/comments, collections
with read-time duplicate checks, numbers incl. the N/M tower question,
symbols/keywords incl. ::auto-resolution, strings/chars), the quote-family
sugars, the full #-dispatch catalog with normative entries (anonymous fn
%-derivation, discard composition, reader conditionals, symbolic floats,
tagged literals), and the syntax-quote contract (core/alias/current-ns
qualification, template-stable gensyms, ~' idiom, distribution through
collections).
Adapting the corpus surfaced and filed three findings, recorded as labeled
divergences/UNVERIFIED in the chapter: nested syntax-quote doesn't collapse
(S25, (= "meow" ```"meow") is false), #inst reads as a bare string (identity
data reader, no instant type), and jolt satisfies :clj in reader
conditionals (feature-key policy under review).
reader-forms-spec gains 11 chapter-cited rows (discard stacking, ##Inf/
##-Inf/##NaN, :default conditionals, qualified var-quote identity, gensym
stability within vs across templates) — all passing.
Fifteen vars from the spec coverage gap (docs/spec/coverage.md):
parse-long/parse-double/parse-boolean (strict validation; scan-number alone
accepts 0x10), newline, current-time-ms (host clock for time), update-keys/
update-vals (PHM base, collisions last-wins), partitionv/partitionv-all/
splitv-at (lazy seqs of vectors; splitv-at's tail stays a seq, matching the
reference), with-redefs/with-redefs-fn (roots restored on throw), time,
macroexpand (expand-1 to fixpoint), alias/ns-unalias (write BOTH alias stores
— require :as uses string-keyed :imports while ns-aliases reads :aliases;
split filed), ns-publics (symbol-keyed map; publics == interns, no privacy).
Three pre-existing bugs fixed along the way:
- (partition n step pad coll) misparsed pad as the coll and returned ()
- (require 'bare.symbol) rejected — only vector specs were accepted
- analyze-form leaked the interpreted analyzer's ns on a punt: a throw out of
the analyzer left current-ns=jolt.analyzer and :compile-ns set, so the
fallback interpretation resolved user vars against the wrong namespace
(bit (var user-sym) under compile mode)
And one overlay-authoring landmine documented in-code: a 20-coll fn must not
use 30-macros macros (with-redefs-fn's dotimes compiled as a forward ref that
resolved to the macro fn at runtime) — loop/recur instead.
Gate: conformance 316x3 (+14 rows), suite 4324 pass / 78 clean (was 4081/72;
parse_*/update_* files now contribute), baselines raised, all specs+unit,
fixpoint, self-host, sci, staged. Coverage: missing-portable 35 -> 20.
RFC 0001 proposes a normative, implementation-independent Clojure language
spec (the reader, evaluation model, special forms, data types, seq/laziness
contracts, namespaces/vars, and the portable clojure.core surface) to the
standard of R7RS/Racket — Clojure has none, and every alternative
implementation re-derives semantics from the reference and folklore. The
spec is executable-first: every numbered normative statement cites its
conformance test or is marked UNVERIFIED.
docs/spec/ carries the front matter (conformance terms, entry format, host
classification), the special-form catalog with worked normative entries for
if and let*, the core-library entry format with worked entries for first,
reduce, and parse-uuid, and a generated coverage dashboard over the 694-var
ClojureDocs inventory (tools/spec_coverage.py cross-references the surface
against jolt's interned+resolvable vars and the test suites).
Measured baseline: 380 implemented+tested, 154 implemented-untested, 35
portable-but-missing (filed), 22 resolvable-but-not-interned (filed — seed
fns invisible to resolve/ns-publics), rest classified host/JVM/concurrency.
