Liking cljdoc? Tell your friends :D

Changelog

0.6.0 — 2026-08-12

What a stored rule is worth. A rule can now conclude a rule, a NAF guard written as a conjunction guards instead of firing unconditionally, and every door that reaches a rule reads belief rather than storage. Beside them, the arrival-order dependences left in the belief loop itself are closed — a revived datum, an un-merged spelling, and every report, digest and election that keyed on retrieval order — and two reads that grew with what the KB holds rather than with what the write touched now read forward off the region: except's visibility set and the planner's subtype fan.

Three entries are Breaking, which is what makes this a minor rather than a point release: the daemon answers an export refusal 400 where it answered 500, the model's tool surface drops two ops, and the CLI's refusals move to stderr. Seven Refusal entries batch here rather than each forcing its own release, which is what §3.8 of CONTRIBUTING.md designates a minor for. Every Breaking and every Refusal entry carries its Migration line.

Triage, for a 0.5.1 caller. This is the index to what touches something you have written.

If your code…Then
treats a 5xx from the daemon's :export op as a backend faultthe five destination refusals answer 400 now; retry logic keyed on 5xx stops retrying a caller mistake
drives :preview or :clear-caches through the LLM tool surfacethey are no longer exposed to a model — call the op on the daemon or the API directly
writes (ist Ctx S) in an antecedent or an exceptWhenrefused :not-well-formed; say it with decontextualizedPredicate or a genlContext edge
passes (ist Ctx S) to a readit answers now instead of returning empty, with the named context winning over the argument
writes (unknown (and A B)) as a guardit guards now; it fired unconditionally before. Under a quantifier the same shape is refused
branches on violations' :violation with a defaultless case:functional, :asymmetric and :constraint-exposure-truncated are new kinds
runs check over (not (implies …))refused :not-well-formed at both doors, where check passed it and assert threw
runs a :refuse KB and reads an empty violations as a clean billcross-context functional and asymmetric pairs are reported there now
passes --strength to the CLI's assert-rule, or reads CLI refusals off stdoutthe flag is honoured now, and refusals print on stderr
forks a KB with an opts map naming neither :space nor :dirthe fork lands on its own space instead of the shared process default
asks a symmetric predicate about a claim that is inherited rather than storedthe mirror composes with the other provers now, so an ask can answer more
writes a kind-level (hasCapability <kind> …) against the shipped ontologykind-level claims are capabilityType; hasCapability is the instance-level reading alone

The shipped ontology now satisfies the declarations it ships. hasCapability was read at two levels by one symbol, and one symbol gets one argument check.

  • A capability claim about a kind is capabilityType, and about a member is hasCapability. (hasCapability bird flying) said the kind flies and (hasCapability Tweety flying) said one bird does, off a single predicate whose first position was declared argIsa … animal — right for the member and wrong for the kind, since bird lies under animal rather than in it. So seven facts the starter shipped were convicted by a declaration the starter also shipped: check reported :arg-type on each and re-asserting any of them threw, while loading them was silent. They loaded because the declaration lives in LifeContext and the facts in BiologyContext, an ordering the open-world reading accepts and for which violations files no retroactive :arg-type report — so nothing said a word. The kind-level content moves to capabilityType, a typeRelationPredicate taking argGenl on both positions and carrying the argPreserving/argPreservingInverse pair that reaches the kinds beneath and the capabilities above; hasCapability keeps argIsa … 1 animal and is the instance-level reading alone. No kind-level travelling rule replaces the six conclusions that vanish with it: argPreservingInverse answers them at retrieval, and nothing at that level rests on a record. Guards: ontology-test asks each reading its own question and checks neither answers the other's, plus two sweeps for the gate that was missing, since loading is not checking and the ordering that admits a sentence is gone by the time anyone reads it. starter-test's every-sentence-the-starter-ships-is-well-formed-once-it-is-all-loaded puts all 1,150 sentences of resources/kb/ to check against the fully loaded KB — the authored side, so it covers the rules, which reach the store as slots and a sentexHandle reference no .txt contains. ontology-test's every-fact-the-starter-ships-satisfies-the-declarations-it-ships walks the stored facts, which is what catches the six a rule derived. Either alone misses what the other finds. Class: none. resources/kb/ is data rather than surface (CONTRIBUTING.md §3.8), so a content edit rides any release; what it owes is the roster that makes it visible, and the sweep above is it. docs/inherit.md, docs/taxonomy.md.
  • The pairing is prose, because the vocabulary cannot state it. typeToInstancePred is the link recording that two predicates are the type-level and instance-level readings of one relation, and it constrains its second argument to an instanceRelationPredicate. hasCapability cannot carry that mark: it relates one animal to a capability kind, and a marked predicate must use one argument-check family for every position, so the mark would cost (argGenl hasCapability 2 capability). relationKind classifies a predicate whose two ends sit at one level — resultIsa and functionCorrespondingPredicate are unmarked for the same reason — so the pair is named in both comments and the enforced check is kept over the inert link.

A guard keyed on the operator instead of on what it reads. Two doors and one index were answering about negation-as-failure conditions without ever being asked again.

  • An exceptWhen whose query is itself a query operator is now watched by what the query reads. The exception is any closed level-6 goal, so (unknown S), a thereExists and an aggregate all stand there — and the re-check index was keyed on the conjunct's own functor, which for those three is a symbol no sentex is ever stored under. The rule sat in the index under a predicate nothing arrives on, so no fact could queue it: the exception was evaluated correctly the first time and re-evaluated never, which is not an exception that fails safe but one that answers whatever happened first — (exceptWhen (unknown S) R) blocked R forever, including after S became derivable and the guard should have released. The stratification graph read the same keys, so a cycle through such an exception reached no rule and was refused nowhere. rules/watched-predicates peels the frames — an unknown to its conjuncts, a thereExists to its body, an aggregate to its census body — and is read by all four sites that key on them (the live registration, its removal half, the index rebuild, and the stratification graph), so a rebuilt index cannot disagree with a live one. not is deliberately not peeled: the trigger side keys an arriving (not S) under not too, so the two agree. Guards: except_recheck_test's release-after-the-fact case, which is the direction an initial answer cannot show, and its stratification twin. Class: neither label — belief moves for a rule of this shape, and no working caller exists to break, the guard having answered from arrival order rather than from content. docs/exceptions.md.
  • And the settle-time narrowing peels the same frames, or fixing the key would have bought a quadratic. The filter that decides which firings to re-evaluate substitutes the firing's bindings into each block literal and asks whether the trigger could answer it; a literal that is not flat and ground has no shape to test and is kept, which is the safe direction and, for an exception that is a query operator, is every firing every time — (unknown (qskip PX7)) reads as unshaped however specific it is. With the rule now correctly queued, that made the re-check cost the rule's whole history per arriving fact: 48 → 192 level-6 evaluations for the same 6 triggers as the firing count went 8 → 32, against 0 once the frames are peeled, the triggers being about individuals no firing binds. Both populations peel — the placed justifications and the refusal record, which is where this shape accumulates, since (unknown S) holds exactly while S is absent and that is the state a firing is refused in. Guard: except_recheck_test's counting test for the query-operator shape, beside the flat one it mirrors — a call count rather than a clock, for the reason that file gives. Class: neither label — a cost, and one no released engine ever paid, the key that would have reached these firings having landed in the same change. docs/exceptions.md.
  • check predicts the NAF-literal refusals instead of only assert throwing them. Closure, quantifier locality, the aggregate reduction slot, a quantified or empty conjunction — all of them live in sentex/check-naf-closed, which the constructor runs, so both storage doors had them and the dry-run door did not: check predicts an assert without building a sentex. A caller validating a rule before writing it was told the rule was admissible and then handed a throw, which is the one answer check must never give. Moved into checks/check-rule!, the list every door reads, so the prediction is made where a prediction can see it. Guard: check_test holds seven shapes to one :type at both doors and to no writes. Class: Additivecheck reports problems where it reported none, which is the promise its docstring already made. docs/naf.md.

A rule can conclude a rule. (implies <antes> (implies <antes'> <conseq'>)) is a generator: its firing stores the rule it concludes, holes filled.

  • The hole split is computed, not declared. A variable the generator's own antecedents also mention is a hole — bound by the join, ground in the mint; every other variable in the stamped rule is that rule's own and survives as a variable. Sharing a name is how an author says "fill this in", so there is no template vocabulary, no holes vector that could disagree with the template it annotates, and no synthetic predicate joining two sentences that have to be kept in step. Range restriction moves one level in accordingly: the generator's own is vacuous — its consequent is a rule, not a conclusion — so what is checked is the stamped rule's, with the holes counted as bound. A hole may stand in functor position, which is the point: one generator ranges over a family of predicates while every rule the index ever keys on has a concrete functor, so the variable-predicate refusal stands untouched and its advice — assert the instantiated rules, one per predicate — stops being manual labour. A stamped variable functor that is not a hole is refused like any other. Class: Additive — a shape that was refused is now accepted. docs/generators.md.
  • A mint is derived content, which is the reason to mint rather than macro-expand. The stamped rule is justified by the firing (antecedent handles plus the generator's), not marked a premise, so retracting or defeating what licensed it un-believes it through the ordinary relabel and it stops firing — no separate un-mint path, and why on a conclusion reads through a real rule handle to the facts that produced it. It is stored through the same check list the assert door runs (checks/check-rule!, now the one reader both doors share, so a check added at one cannot be missing at the other) and polycanonicalized the same way; a mint that cannot stand is dropped and recorded rather than thrown, since a fixpoint may not abort halfway through itself. Both arrival orders agree with no sweep of the generator's own: a generator is an ordinary rule, and a newly asserted or newly minted rule is a datum that joins over what is already stored.
  • Rules now follow belief, and that is what makes the above work. Both chainers reached a rule through the rule index, which posts on storage and knows nothing of belief, so a rule whose support had gone went on firing forward and answering backward goals. res/rule-believed? is asked at all four sites (chain/fire-rules-for, chain/process-datum, and both candidate-rules). A sentex the TMS holds no node for is available, not disbelieved — an :inert rule is stored outside belief on purpose — so the arm reads an absence as an absence. No shipped behaviour moves: until now nothing could make a stored rule un-believed. Class: Additive, for that reason.
  • One level of nesting, and five refusals. A rule generating a generator is :not-well-formed — the scoping rule needs one split per sentence, and a second would have nothing in the spelling to disambiguate it. An exceptWhen on the stamped rule is refused the same way, and this one is a silence rather than a nesting: an exception is a meta-sentex keyed by the rule's handle, split off by the assert path, which a firing does not run — so the mint would be a rule whose guard had evaporated, firing on exactly the bindings its author wrote it not to. An (unknown …) antecedent inside the stamped rule, or an exceptWhen on the generator, both work and the message names them. A set/backwardRule generator is :not-indexable: no backward goal asks for a rule, so it would claim a capability it cannot exercise. A generator sharing no variable with the rule it stamps is :not-range-restricted — every firing would stamp the same rule. And a generator cycle — a stamped rule concluding a predicate some generator reads — is :not-stratified, refused outright rather than depth-capped, because a cap makes the KB's contents a function of how long the chainer ran; the check runs both directions at every generator's assert, or the cycle would be admitted whenever the two were asserted in the other order. check predicts every one of them.

A NAF guard written as a conjunction now guards. (unknown (and A B)) is the exceptWhen conjunction inlined per literal, and reads the same way: block when every conjunct is derivable.

  • The form was accepted and inert, in three places at once. No prover claims the functor and, so chain/unknown-inner-holds? handed the whole (and A B) to the level-6 registry as one goal, got no answer, and read "not derivable" — which is the unknown holding, so the rule fired unconditionally. The two bookkeeping reads were blind the same way and for the same reason: rules/naf-predicates posted the rule in the re-check index under the predicate and, which no fact ever carries, so no arriving datum could ever queue it; and checks/negative-predicates drew its stratification edge from that same functor, so a cycle through such an unknown was invisible to the check that exists to refuse one. An author who wrote a two-condition guard got a rule with no guard at all, and nothing said so. sentex/naf-query-conjuncts is the one accessor all three now read — the unknown spelling of exception-query-conjuncts — so the query's conjuncts are evaluated by provers/exception-holds? (block-if-all, the exception's own evaluator, which is what keeps the two from drifting), every conjunct's predicate is watched by the re-check index, and every one is a negative edge. UnknownProver reads the same accessor, so the goal (unknown (and A B)) and the antecedent agree. Neither order nor nesting nor repetition is the rule's identity: the conjuncts are sorted as an exception's are, and flattened, because a nested and is a goal no prover claims either — one left whole would come back unanswerable and read as not derivable, which is the same silent hold one layer down. A conjunct may itself be a thereExists, its binder being local to it, and the predicate watched is the one inside the quantifier. Guards: naf_test's eleven new cases, including the arrival on the second conjunct's predicate, which is what the old index could not see, and the stratification cycle closed through that same conjunct, which is what the old negative edge could not see. Class: neither label — belief moves for a stored rule of this shape, but no working caller exists to break: the guard never guarded, so no author's code was doing what its author believed. docs/naf.md.
  • A conjunction under a quantifier is refused(unknown (thereExists ?c (and …))) and the aggregate body that spells the same shape, both :type :quantified-conjunction. There the conjuncts share the binder, and the registry answers one goal at a time: read flat, each conjunct would be satisfied by a different witness, so "has a sick child" would hold of anyone with a child while anyone at all was sick. Accepting it stores a guard that convicts on evidence about two unrelated individuals, which is worse than the inert one it replaces — so sentex/check-naf-closed refuses it at the door, naming the repair (bind the witness with a generator antecedent). The existential unknown is therefore one literal deep, which is what it always evaluated as. An empty conjunction is refused beside it as :not-well-formed: nothing can make it derivable, so the antecedent guards nothing. Class: Refusal — no working caller exists, both shapes having silently held rather than blocked; the new :type keyword is Additive. Migration: bind the witness with a generator antecedent and leave one literal under the unknown; a KB that asserted either shape was storing a guard that never blocked. docs/naf.md, docs/aggregate.md.

The strictest policy stops being the leakiest. Under :refuse a cross-context functional or asymmetric clash was neither refused nor reported; both are now ledger entries, beside the :disjoint one that already was.

  • Two of the three clash kinds slipped through exactly where nothing was meant to. The definitional checks are scoped to the writer's own cone, so a pair split across a genlContext edge is invisible to both writers and the assert door refuses nothing. Under :arbitrate that is answered by the vantages — a context that sees the pair whole weighs it — but under :refuse the vantages are deliberately withheld, and the exposure ledger had an entry kind for disjointness only. So a functional slot filled either side of the edge, and an asymmetric claim written across one, stood believed and unmentioned in the policy chosen precisely to let nothing through. settle/expose-constraint-clashes! is the reporting half: :functional and :asymmetric entries carrying :pred, the two [sentence context] halves as :clash, and the :visible-from vantage, shaped like the :disjoint entry beside them. It reports and never decides — belief is untouched under :refuse by construction, contradictions stays the answer to what was arbitrated, and a pair this settle arbitrated is excluded exactly as the disjointness pass excludes one. Guards: exposure_test's twelve new cases, including the :arbitrate counterpart that catches the gate applied in the wrong place, and the same-context case that must still refuse rather than report. Class: Additive — a new entry kind in an accumulating ledger, and both renderers read (name (:violation v)) rather than dispatching on a closed set. A :refuse KB that saw an empty violations may now see entries, which is the point rather than a migration. docs/nmtms.md.
  • The discovery is the refusal's, asked from somewhere else. The pass re-derives each clash through checks/arbitrable-violations from the vantages clash-vantages names — the same two reads clash-askers makes under :arbitrate — so the report cannot drift from what the door would refuse, and closing the gap widened no vantage. The pair was always visible from there; what was missing was an entry kind to say so.
  • It costs a KB that declares neither property two seqs. The pass is :refuse-only and gated before any root is read, behind an O(1) check on the declared vocabulary (tax/props :functional / :asymmetric). Its candidates are the moved region's own binary facts rather than a declaration's reach, so on a plain assert it works out no extent — which also means a (functional P) arriving after the facts it convicts is the arbitrating sweep's question and not this pass's, an absence docs/taxonomy.md states rather than implies.
  • A genlContext edge is the one trigger that reaches past the region, and it has to. Visibility itself moves there: a pair whose halves are both stored and both believed becomes jointly visible with neither half relabelled, so taking candidates from the moved region alone reported the same knowledge when the edges arrived before the facts and never when they arrived after — the arrival-order dependence the pass exists to remove, showing up in the pass itself. An edge in the region reaches out over the cone it newly sees (constraint-facts-in-cone, the binary-fact parallel of the disjointness pass's members-in-cone, spending the same *exposure-instance-budget*), so both passes answer that trigger the same way. Below the cap a cone walk is proportional to the cone, deliberately; the checkable property is flatness past it, and perf's constraint-exposure-context-edge reads 0.89x across 8x the facts behind one edge at a budget of 100. It failed on its first run at 5.42x, against a claim that was what was wrong.
  • And a KB that declares one costs what the region holds, which took a second pass to make true. The vantage search read the argument-1 posting of both arguments of a candidate, where only one can hold a partner — a functional partner shares argument 1, an asymmetric one holds it at argument 2 — and on a term shared across an extent the useless posting is the extent. Asserting into a declared-asymmetric slot sharing argument 1 grew 2.88x from 250 to 2000 facts against a flat 0.91x with the pass off, which is 3.9x per assert at the top end. settle/partner-contexts now reads the one posting each declared property could hold a partner in, and the same load reads 0.89x. The narrowing is on the shared path, so the :arbitrate vantages stop making the same useless read. perf's constraint-exposure-shared-arg is the gate and is the only check in that file whose KB declares a predicate property — every other builds with fresh-kb, which declares none, so the pass shuts at its vocabulary gate and the suite could not see it at all. It failed on its first run.
  • Entries are capped, and never silently. The candidates are the region but the entries are not: one slot filled from N contexts a single vantage sees is N−1 pairs off one arriving fact, against a ledger that keeps 1000. One pass files at most settle/*exposure-instance-budget* and files a :constraint-exposure-truncated entry when it had more. A separate kind from the two sweep notices for the reason they are separate from each other: a reader acts differently on these pairs went unreported than on this trigger went unswept. The same entry says how many genlContext edges went unswept, with a sample, a cone walk that stopped being the same class of thing to a reader — pairs visible and unreported, nothing left undecided.
  • One entry per pair, keyed on content. Both halves can sit in one settle's region and each convicts the other, so a report keyed on the walked side would file a pair twice — or read differently — depending on which arrived last. The entry is keyed on the handle pair, the halves are ordered by printed form, and the top-level :sentence / :context name the first of that ordered pair rather than the side that found it. Guard: exposure_test/the-report-is-the-same-in-either-arrival-order.

A hidden set kept where the sentexes are, not rebuilt per placement. except's visibility read cost as much as the KB hides, every time it was asked — which is once per placement and once per candidate justification. It reads a roster maintained at the store now, and is flat in what the KB hides.

  • The read was linear in the number of excepts and is now flat in it. res/excepted-handles answers which handles a believed (except (sentexHandle H)) hides from a view context, and it fetched a record, re-derived a target and asked jtms/in? for every stored except in the KB per call. kb/note-excepted! maintains {context -> {hidden-handle -> #{except-handle}}} at create-sentex and integrate/sentex-removed! instead — beside the :opposed coincidence set, on the same terms and for the same reason, and rebuilt by recover because it is derived from storage and no store holds it. On a chaining run of 400 facts and 380 derivations (lein bench-hotreads, best of 3):

    exceptsµs/derivationns/readshare of the run
    0237.0 → 243.0216 → 1250.7% → 0.4%
    10184.4 → 165.33,562 → 59412.0% → 2.5%
    100380.4 → 156.325,936 → 63856.3% → 2.5%
    1,0002,294.1 → 143.6185,458 → 55388.8% → 3.1%

    E = 0 is unmoved, which is the second claim: a KB that hides nothing still pays one question and gets an empty answer. Class: neither label — res/excepted-handles keeps its arity, its contract and its answer set; what moved is where it reads them from. docs/contexts.md.

  • Belief stays a read, and the roster is storage only. An except can be defeated or revived with no sentex arriving or leaving, so there is no choke point a believed-set could be maintained at — the roster holds what is stored, and whoever reads it filters by belief, which is the line :opposed draws. It is also why this is a roster rather than a clock-stamped memo: the scope that asks is forward chaining, which writes while it reads, so a stamped entry would be retired between one placement and the next.

  • The callers with handles in hand stopped materializing a set to look one up. res/hidden-fn hands back a predicate over one view context — nil when that vantage hides nothing, so a caller skips its filter rather than running one that can only answer false. without-excepted, chain/antecedent-hidden? and kb/types-of take it; excepted-handles remains for the caller that wants every hidden handle. Building the set costs one pass over the reader's excepts however few handles will be asked about, and the questions are bounded by the answer set while the excepts are not.

  • The gate is the roster being empty, which is tighter than the count it replaces — a KB storing only (not (except H)) roots under except and counts non-zero. Six assert_cost_test budgets are re-pinned for the index reads that gate no longer makes, every one of them downward.

  • A roster that drifts is a wrong belief, not a slow one, so the guard is an oracle rather than a behaviour test: meta_sentex_test compares it against a full scan of storage after every kind of arrival, defeat, revival and removal, across a recover, and compares hidden-fn against the set read for every handle either could be asked about.

  • The planner's subtype fan is made cheap rather than remembered. est-matches costs a unary type literal over the type's whole subtype closure — the one branch that is not a handful of O(1) index reads, asked once per pick, per plan, per firing attempt, and 13.2% of a forward chaining run on a 364-type hierarchy. For the shape that costs, the argument a bare open variable, the general walk is provably a long way round to count-at [t']: the literal's token stream is the functor, which extends the trie prefix, and then the variable, which stops the walk. fan-of-roots reads those counts directly and the fan halves, to 6.8%. Any deeper prefix — a compound argument, or a partly-bound one — still takes the general walk, because for those the prefix genuinely is deeper. Class: neither label; the number returned is identical by construction, and plan_test pins it against the walk itself rather than against a number written down in a test. docs/inference.md.

  • Remembering that answer instead does not work, and the harness says so on both paths. A memo stamped on the change clock measures 0.98–0.99× under chaining, because the run's own placements retire the entry between one plan and the next; on a query, where nothing moves the clock and every plan after the first would be served, the fan is under 3% of the run and it measures 0.91–1.04×. A finer stamp is unsound rather than merely fiddly — the estimate bounds from above, a reading of 1 is a proof rank-blocks and cartesian-factors rest on, so an entry computed before a placement is too small. lein bench-hotreads's third arm reports both paths, so the next reader inherits the measurement rather than the idea.

  • lein perf gains the check that would have caught the visibility walk. The read is now flat in what a KB hides, and flatness is a growth claim a ratio can see — which the old shape was not, being linear in it. visibility-reading times a scoped read at 8 and 1,024 excepts, all of them hiding decoys the read never returns, so n moves the filter's input and leaves its output alone. Calibrated from both ends as this file requires: 0.91× and 0.90× on full runs, against 27.33× measured by running the check against a tree that re-derives the hidden set per call. The bound is the flat claim's own 2.0×. The literal cache is bound off inside it, because a repeat read under an unmoved clock is served whole from that cache and would report the filter free at both sizes — while the caller the check exists for, forward chaining, moves the clock per placement and meets this read cold every time.

ist places, and four layers had it half-reading. A rule cannot qualify a premise by the context to read it from, and now says so at the door instead of storing a rule that decides itself on a context it never consulted.

  • An (ist Ctx S) in antecedent or exceptWhen position is refused as :not-well-formed. The literal is indexed and matched under the functor ist (rules/antecedent-predicates, res/match-pattern), which no sentex carries, so it satisfies nothing and no arriving datum triggers it — while four layers around it read the frame as meaningful: the naming check descends the context slot and refuses a non-context there, range restriction counts the slot's variables as bound by an antecedent that will never bind them, canonicalization sorts the literal by its inner predicate, and well-formedness accepted it in every role. So check reported no problems and assert returned a handle. What accepting it does to the store is the argument for the class: a positive antecedent yields a stored rule that cannot fire; an exceptWhen query never matches, so the guard never guards and the conclusion it was written to block stands believed; and an (unknown (ist …)) is satisfied by that same emptiness, so the rule fires unconditionally. A rule that does nothing announces itself — the other two do not, which is why this is refused rather than left inert. The refusal names the two ways to say what such an author meant, since a violation reported without its repair is a second lookup: (decontextualizedPredicate P) takes every (P ...) into UniverseContext, which every context sees, and a genlContext edge puts Ctx in the rule's own cone — under either the premise is written plainly, and the context topology decides what is readable from where. The NAF frame is now descended by the connective walk in its literal's own role, which is what carries the refusal into (unknown …); a top-level one is not descended, unknown being an antecedent/exception construct. Guard: check_test's five frames, holding both doors to one :type and the store to no writes. Class: Refusal — no working caller exists, the shape having never matched anything. Migration: say it with (decontextualizedPredicate P) or a genlContext edge into the rule's own cone — the refusal names both. docs/contexts.md.
  • And it reads on a caller's behalf, which is the half that was missing. (ist Ctx S) handed to a read returned nothing at all — the most natural thing to type against a context, answering as though the context were empty. Every read taking a sentence and a context now takes one, asking S in Ctx with the named context winning over the argument, which is the resolution assert already makes: one form, one meaning on both sides of the KB. That is sentexes-matching, handle-of, ask, prove, query, query-plan, ask-within, prove-within, why-not's sentence arity and the three level diagnostics — the rule being the surface, so there is no list to learn. contexts-of and find-sentexes take no context and ask which contexts hold a sentence, so the form is not a question they have; isa? / genls take a context but a term rather than a sentence. Each door answers at its own notion of a context, and the two families differ on purpose: retrieval returns the sentexes stored in Ctx, while the reasoning doors answer from everything Ctx inherits. Both are "in Ctx", a fact Ctx inherits being true in Ctx, so the difference is the doors' and not ist's — pinned rather than reconciled. This grants no visibility a context argument did not already grant, which is what separates it from the antecedent above: naming AContext is what (sentexes-matching kb S 'AContext) has always done, and the caller asking has said so, where a rule reading Ctx on the sly would decide belief from a context its own cannot see. Guards: context_scoping_test's four, including the whole-surface one and the sibling lattice where the answer is a context the asker cannot see. Class: Additive — a goal shape that returned empty now answers, and no read's behavior on any other input moves. docs/api.md, docs/contexts.md.
  • Two read shapes are refused instead of answered empty. A wrong-arity (ist Ctx) or (ist Ctx S junk) is assert's own :shape — the same refusal on the read doors that assert and check already made, so a caller cannot get a silent empty from a form the write door rejects. And an (ist …) standing as a conjunct of a vector goal is :not-well-formed: a join's conjuncts share their bindings, so there is no per-literal context to honor, and it is the antecedent question wearing another frame. What accepting them does is the argument for the class — both return an empty result set, which is indistinguishable from a true negative, so the caller reads "nothing holds" where the engine meant "I did not understand the question". Class: Refusal — no working caller exists, both shapes having answered nothing. Migration: nothing; both forms returned an empty result set, so no answer a caller held moves. docs/api.md.

A datum that comes back believed goes back on the agenda. Two routes let belief arrive with nothing chaining behind it, so the same knowledge in one order concluded and in the other did not — the invariant the README states first, failing in the loop that exists to hold it.

  • A revived antecedent licenses the firing its defeat withheld. chain/*matcher* is belief-filtered, so an OUT datum is not a match and the join yields no candidate; reviving it yielded none either, because the firing that never happened left no justification for a blocked set to release and reached no placement for the refusal record to re-ask. Both existing instruments are blind to it by construction. So the trigger is read where the belief moved: jtms/revived, with settle re-seeding those datums onto the chaining agenda. The cost half is jtms/touched-new, a third window set naming the nodes the window created — without it every asserted fact and every conclusion drawn from one reads as newly believed, and the re-seed becomes a second forward chain per settle. The distinction exists only at creation, so both TMS representations take it there. A rebuild stands aside: recover relabels the whole graph and replays justifications that already carry what was derived. Class: neither label — belief moves only toward conclusions the same knowledge already reached in another order, which is not a contract a caller could hold. Guards: revived_datum_test and twenty orderings in order_independence_test, six of which disagreed. docs/nmtms.md.
  • The equality door of the same defect, and jtms/revived cannot see it. A merge displaces a spelling and its twin joins in its place, so a partner arriving mid-merge concludes at the twin; stop believing the equality and the twin is swept while the displaced spelling comes back — leaving both antecedents of a forward rule believed and neither spelling of their conclusion held. Supersession moves belief with no relabel behind it, deliberately (a superseded datum stays :in so its twin's justification survives), so the flip is in none of the three window sets. refresh-supersessions is where the answer exists — settle-finish already brackets it — so the spellings it gives back go into *unmerged-sink* and settle re-seeds and settles again, bounded by max-unmerge-rounds, the way retract! already settles twice around its own re-derivation. Two alternative designs lose on what they cost elsewhere and docs/nmtms.md says why; the route is not reachable by negating the equality, since the merge rewrites the negation's own terms. Class: neither label, on the entry above's argument. Guards: both un-merge routes — a retracted equality and a functional merge losing a filler — plus twenty orderings, six of which disagreed, and a cost guard on rechain-seeds whose merging arm requires that displacing a spelling puts nothing on the agenda. docs/equality.md.

An answer picked from a fan is keyed on content, never on arrival. Fourteen reads elected a survivor, a representative or a display line by retrieval order, which under the columnar index is assertion order — so the same knowledge answered differently depending on how it was loaded.

  • Two keys were being elided by an ambient print setting. solve/content-key and skolem/rule-digest bind the print vars off: a caller's *print-length* cut the content — and content-key's last-resort handle — out of a key that decides arbitration and out of a digest stored durably in termOfUnit content, felling both back to arrival order. The same binding is now made wherever EDN is written for something other than a human to read: the browser's editor seed line, diff-order, the proposal panel's hidden line and the LLM selection's edit line, since elided EDN is legal EDN naming something else — a saved but untouched editor panel would retract the real fact and store the mutilated one.
  • The elections themselves. dedup-constant resolves a colliding expression to the survivor group-collisions elects; a clash report sorts each side's justifications by the key core/supporting-justifications reads through; one-supporter breaks a same-context tie on the printed sentence; a term commented in two contexts glosses from the content-least; rewrite-target answers nil unless exactly one believed rewriteOf names the expression, as correspondence-of already did; why-not's :contradicted-by and excepted-argument's completion are content-ordered; quality's rule line adds the context to its tie key, one implication in two contexts having shared sentence and signature ahead of the display cap; strongest-per-tuple and support-for break class ties on printed form after the context, the matcher being type-aware; edge/translate sorts nogoods by member content before emission, so two arrival orders render one ASPIF text; and label-context mints its ist copies in content order, as label-dilemmas does. core/supporting-justifications answers in content order at the source — informant's sentence, then antecedent sentences — which is what preview's named reason, why's :support and a clash report's :justifications all read through.
  • And the handle cache stopped answering from another KB. canon-stamp carries the record store beside the symmetric set and stamps compare with =: with-handle-cache reuses an outer run's map, and two KBs declaring nothing symmetric stamped the one shared empty set, so a nested run on a second KB read the first KB's handles. find-sentex-handle caches only a spelling canonicalization leaves alone, the removal choke point clearing the canonical key — a mirrored or folded spelling's entry outlived its sentex as a stale handle. asymmetry-problems compares the canonical converse, so a folded comparison predicate's opposing sentexes are found at all. Class: neither label for all three bullets — every read answers the same set, in an order that is now a function of the content; the previous order was not reproducible for the same knowledge, so there was nothing stable to depend on.

A collected NAT leaves none of its bookkeeping behind. nat/bookkeeping-handles answered lazily and the caller retracts what it hands back, so the answer was being computed against a KB the loop had already torn pieces out of: whether a sentex is one of k's own is read off k's termOfUnit, and a tail forced after that map's own retraction finds no expression. The materialized result types and the correspondence projection behind it therefore stopped looking like bookkeeping and stayed stored — a raw nat/ symbol left naming a constant the sweep had collected, in the retrieval orders that hand back the map first and not in the others. The set and the sweep's own orphan list are both realized before the first retraction now. Class: neither label; what it removes is bookkeeping for a constant that no longer exists.

A stored sentex is not a believed one, and five reads had it the wrong way round. Beside the rule index above, four reads took storage for belief and one took belief for storage.

  • ASP grounding takes only believed assumption and constraint rules; a superseded rule minted choice heads and went on forbidding models. The {:belief? false} import stores each record with its dump strength and premise mark, so a later recover has premises to believe rather than a store where nothing grounds. The catalog's belief caveat probes for a believed datum rather than for any node, which a recover over a strengthless store builds per handle with all of them OUT. The generator reports :stored as storage rather than as a sum over the believed context closure.
  • The converse correction is the reified-NAT sweep: uses count by storage, since a stored-but-OUT use revives and an inert choice head has no node, so collecting the map from under either dangles the constant and re-reifying mints a second one. The teardown list stops belief-filtering for the same reason, and docs/nat.md says stored where it said believed. Class: neither label — each read now answers the question its docstring already claimed.

Retrieval answers what the reference answers. Four matching reads disagreed with the fan-out they are checked against, three of them silently.

  • The mirror probe asks the candidate's own functor. matches-hierarchical mirrored a candidate whenever some predicate under the queried one was symmetric, so with a symmetric sub-predicate anywhere in the hierarchy a stored (knows Bob Ann) answered (knows ?x Bob) — an extra answer the reference fan-out refuses, on the default retrieval path. Class: neither label; the answer withdrawn was one the two paths disagreed about, and docs/inference.md names the fan-out as the arbiter.
  • naf-query unwraps an aggregate as it unwraps thereExists, so a rule with an unknown-over-aggregate antecedent re-checks on the body's predicate rather than on agg/count, which no fact carries: the count moving no longer leaves the old conclusion believed, and stratification sees the negative edge. The same shape as the and guard above, arriving through the aggregate frame. docs/naf.md, docs/aggregate.md.
  • Three spellings stop being three sentences. The rete alpha matcher skips exceptWhen meta-sentexes as res/match-one does; aggregate-values normalizes compound values through representative-term, so two spellings of one merged measure count once; and the three pre-canon reads that gated on the list spelling take the vector spelling as the same sentence, which canonicalization has always made it — the NAT leaf test reifies a vector-spelled application instead of storing the raw compound beside the constant, the editor line's ist guard catches a bracketed ist rather than letting it file into the model's context, and the reified-constant expansion walk descends the antecedent vector a rule's record keeps. docs/canonicalization.md.

A symmetric or inverse reading composes with what is derived. The mirror answered off storage alone, so a claim that was inherited rather than stored had no mirror at all.

  • The mirror delegates instead of reading storage. (pred b a) goes back through the registry minus SymmetricProver, so the mirror of a preserved, inherited or computed claim is an answer; *mirror-depth* bounds the re-entry at two levels and falls back to the raw stored read past it. Class: neither label — answers are added, none withdrawn. docs/inference.md.
  • A partner declared on a sub-predicate is the same edge. tax/inverses-under reads the partners of p and of everything under it, consulting the spec closure only where an inverse exists at all; the self-pair arm says what it answers. docs/taxonomy.md.
  • The mirror licenses the forward door too, and the firing names the symmetry it read through, so retracting the (symmetric …) withdraws what only the mirror licensed. The declaration named is the stored sentence's functor's, the matcher mirroring each fanned literal on its own. docs/inherit.md.
  • A defeat inside arbitration re-joins what its sentence licensed, over closures refreshed to what is believed now. Belief flips with nothing arriving or leaving, so nothing queues the re-join an arrival would, and the rest of the settle would otherwise walk a closure still holding the defeated edge. docs/inherit.md, docs/nmtms.md.

A negated rule is refused at the door. (not (implies …)), bare or wrapped, built a RuleSentex whose key cannot be computed — so check answered admissible and assert then threw a bare IndexOutOfBoundsException from inside the store.

  • connective-problems refuses the form as :not-well-formed at both doors, a :type a caller can discriminate on where a JVM exception was neither catchable by kind nor predictable by check. Class: Refusal — no working caller exists: the shape never reached storage, it reached a stack trace. Migration: nothing; assert it as the positive rule with the negation in the consequent, which is what could be stored all along.
  • And a rule cannot be stored inert. assert-inert is the labeling primitive — a recorded truth value, never a claim about the base KB — and a labeling labels atoms, so nothing it exists for wants a rule. What one bought was a rule sentex that had never been through index-rule-sentex, since that runs where a rule is created: no chainer could reach it, and nothing afterwards could, an assert of the same rule resolving to the stored sentex and taking the branch that does not index. The state on the other side was a rule in? called believed and no fact ever fired — the accepted-and-inert shape check-generator refuses at the other door, with the same :not-indexable. The message names the other inertness, which is what a caller landing here usually wants: set/inertRule is a rule that is believed, indexed and browsable and fires neither way — the spelling for a rule kept as documentation, a transitivity the cached closure computes instead (taxonomy.md). This door's inertness is the sentex's: not a premise, so never believed at all. One word for two states, so the distinction is now stated where each is defined — a glossary entry covering both (and recording that there is no :inert strength, the assertable classes being :default and :monotonic), inference.md on why indexing an inert rule is the point, taxonomy.md on writing the transitivity down rather than leaving the KB's most important rule unwritten, and solving.md on why this door refuses one. res/rule-believed? justified its absent-node arm by naming assert-inert and arguing from the :inert direction, two different things, and says what it means now. Guards: an inert rule is believed, :default, and posted under both its antecedent and its consequent predicates — the browsability half nothing pinned — and the transitivity pattern runs end to end, the closure answering genl? / provable? / ask while the rule materializes no edge of its own. Class: Refusal — every caller of this door materializes a labeling's atoms and their negations, which are untouched. Migration: a rule meant as documentation is set/inertRule (or {:direction :inert}) through assert / assert-rule; a rule already stored inert is one nothing was firing, so retract! the handle and assert it.
  • Beside it, five legal API calls stop failing under clojure.spec instrumentation: genl?, representative, same-class?, equiv-class and deprecated? carry their context arity in the spec, and term-role's :ret admits :lexeme and :sense. The dead :arity declaration row is gone, with declared-arity's unused reader parameter and the comment claiming otherwise, and mint-nat!'s docstring says which of the three asserts chain (termOfUnit never does).

Storage keeps no dead frames, and a torn dump refuses. A round of durability fixes across the disk store, the overlay and both import paths — the class the gate's memory backend cannot see, which is why the matrix exists.

  • :truncated-dump (rostered): a dump stream that ends early reads as a clean EOF, so the frames read are counted and compared against what meta.edn states, on both the belief and records-only paths — the comparison replay-index! already makes for index entries. The records-only import likewise refuses a dump naming a handle twice, a BitSet per handle so the streaming path stays streaming; the second frame silently destroyed the first record and counted both. Class: Refusal — no working caller exists: one input is a truncated file, the other loses a record it reports as loaded. Migration: nothing; re-export the dump. Both refusals name what was read against what was promised.
  • No frame whose only fate is a tombstone. unmark-premise! re-stores only a record that carries a strength, on the disk store and the overlay alike (which also stops materializing an override for a derived base record), and the overlay's set insert removes a removal record only when one exists, so a durable fork stops paying two WAL frames per posting. The memory store marks a premise only for a stored record, so premise-ids agrees with the disk store for the same call sequence.
  • And two reads at open. validate-idx-tail! rides scan-idx!'s chunked walk instead of paying one seek and one buffer per slot across the whole idx on every open of every kind; rebuild-premises! tombstones crash damage only (:damaged-dictionary, :malformed-record) and rethrows :unknown-frame, because a build that cannot read a log must not delete it. The fork projection reads entries with first rather than key — the tiered backend yields plain vectors where the map-backed ones yield MapEntrys, so a dense overlay half threw ClassCastException on any export of the fork — and index entries are normalized to vectors at the export frame, nippy freezing the two differently, so a byte-identical logical index stops dumping byte-differently per backend. The dead :base-highest field is gone.

The uberjar loads the ontology it ships. Layer discovery listed a directory, which a packaged jar need not carry, so an uberjar started with CoreContext alone and no upper or middle layer — silently, the KB simply being smaller than the same tree run from source.

  • Discovery lists the jar's own entries, anchored on kb/CoreContext.txt when the jar carries no directory entry, and an unlistable protocol is refused rather than answered nil. Class: neither label — a caller running from source saw every layer already, and one running the uberjar was getting a KB nobody meant to ship. Beside it: the load generator drops a band past the vocabulary instead of drawing from an empty one (predicates 2 under layers 3 put (nth [] 0) inside the lazy rule draw, mid-load, after the vocabulary had been asserted), and refuted-pairs drops retired spellings with res/retired-for? as the positive read drops them through without-retired, a reader below a merge otherwise carrying one negative constraint under two names it knows denote one thing.

ASP routes to a solver that can actually run. AUTO handed off past the size cutoff on available? alone, so a machine carrying libclingo and no clasp binary solved small programs and threw on large ones — the failure arriving with the workload rather than at the probe.

  • The handoff is gated on a once-per-JVM probe that the binary runs, which the facade's available? now reads rather than forking clasp --version per ask. A missing binary is :solver-unavailable: shell/sh execs directly, so the failure is an IOException and the exit-127 arm could never run. clingo's aspif temp file is deleted on any exit, control_new failures included, instead of leaking to deleteOnExit's never-collected hook. dense-roots' reserved family 2 throws :reserved-family (rostered in type_contract_test) instead of pinning a three-element decode for a key whose real shape is four, the packed long having no room for the predicate. Class: Refusal for the two new :types — no working caller exists, both paths having thrown or mis-decoded. Migration: nothing; install clasp if you were relying on the large-program path, which was throwing.

The daemon answers a caller's mistake with 400, and the model's tool surface loses two ops. Both are contract changes a working caller observes, and both are why this release is a minor.

  • :export's destination refusals are client errors. :no-destination, :not-a-directory, :not-empty, :export-busy and :unsupported-format join serve/client-error-types, so a directory that exists and is not empty stops counting as a backend fault at every reverse proxy and 5xx alarm between the caller and the daemon. :sentence leaves the set, a type nothing throws. Class: Breaking — a status code, §3.8's own example. Migration: a client that retries on 5xx and reports 4xx will now report these instead of retrying, which is the intent; wire_contract_test pins the pairing.
  • :preview and :clear-caches are excluded from the tool surface a model reaches. :preview stores nothing but holds the single writer and advances the handle counter, and :clear-caches resets process measurement state — neither is a read, whatever its name suggests. Class: Breaking — the exposed tool set is (keys serve/ops) minus a roster, and this removes two from it. Migration: call either op on the daemon or through vaelii.core directly; nothing about the ops themselves moves.
  • A cancel cannot unsettle a finished job. jobs/cancel! writes :cancelling through update-job! guarded on the job not having settled: stamped over a just-filed :done, the job never settled, the sweep kept it, the writer kept naming it, and every later writing job was refused against work that finished long before.

The browser writes what it shows, and shows what the ledger holds. Four rendering and lifecycle defects, each one a renderer or a guard assuming a shape the data does not have.

  • A violation about no sentence is not a link to a term called nil. Four ledger kinds are about a pair or a budget rather than a dropped sentence, so they carry a :detail and no :sentence; both renderers read those fields unconditionally and term-link's fallback links whatever it is handed, so each row printed "nil" beside a live link to /term?q=nil. The mirror image beside it: :message sits at the top level on :non-confluent and :aggregate where every other kind puts it under :detail, and both renderers read only the :detail spelling, so the one line those entries exist to print was replaced by the generic fallback. Both read either now, and core/violations states plainly that :sentence and :context are not on every entry — the roster this exposed as incomplete (:constraint-exposure-truncated had reached the changelog and nothing else) is stated in all four places. The standing-clash row had the same defect from the other end, handing each [type context] pair to term-link whole.
  • The editor survives a conjunction line, and an export holds the write doors. edit-post keeps the positional pairing per entry and lets a conjunction-concluding line (a vector of handles, assert-shaped) fall to unpaired, where v/sentex refused it :bad-handle after the write had landed and reported failure for a KB that had changed. write-refusal refuses while an export walks the KB (catalog/exporting-kb?, by identity, so a load of another KB is untouched) and the export job drains the write monitor before the walk; unload! waits for a :cancelling loader exactly as for a :running one, the retry the still-stopping refusal asks for having skipped the guard and cleared the stores under the live writer; and a repeat source's key suffix is one past the highest still loaded rather than a count, so unloading generated#1 of two no longer collides the next load with the live generated#2. The caches page and docs/web.md say what the clear does — this KB's entries alone, counters left running — instead of promising a process-wide zeroing the handler cannot ask for.
  • A fork with an opts map lands on its own space. Fork opts naming neither :space nor :dir merge over fresh-overlay-opts: a non-empty map without one sent the fork's writable half to the shared process default, space 0, where two such forks saw each other's writes and a plain open-kb saw both. Naming a space explicitly is still the remount. Class: neither label — the shape it replaces had two forks writing over one another, which no caller can have been depending on. docs/overlay.md.
  • A KB whose store cannot be counted renders :unreadable, not as a healthy empty one (active-caveat folded the throw into zero); a broken foreign reader logs :warn with the cause, and only a genuinely absent plugin stays :debug, the old line having sent operators to reinstall what was already installed; the torn-snapshot refusal carries :entries/:expected in its ex-data as its dictionary sibling does; and a portfolio racer that throws cancels the other racers instead of leaving complete searches running for nobody.

CLI flags mean what they say. Four parsing defects, one of which stored knowledge at a strength the caller did not ask for.

  • assert-rule passes the --strength it parsed — accepted-and-dropped stored a known-true rule at :default, so the record carried a class the caller had not asked for and defeat-class answered it. What that class governs is narrow, and worth stating so the fix is not read as more: it is the rule's own, nothing in the engine defeats a rule (a negated rule is refused, and a dilemma is a fact-level thing — nmtms.md), and what a firing confers is :defeasible's to say. Asserting a rule that is already stored now marks the premise, which is the same door reading the same option and matters more than the class does: a generator's stamped rule is a conclusion resting on the generator, so asserting it returned a handle for a rule that retracting the generator took away — the assertion bought no ground of its own. The class resolves from content, as the two slots beside it do (:direction, :defeasiblecanonicalization.md): the stronger of the assertions stands, since a re-assert carrying no :strength states nothing about the class, and reading that silence as a downgrade left defeat-class answering differently for the same two assertions in the two orders. Migration: narrowing a rule's class is retract! and re-assert, exactly as it already is for direction and defeasibility. A value flag refuses a following flag as its value (:shape), instead of opening a directory literally named --starter and loading no schema. A flag belongs to the commands that read it, and one carried elsewhere is refused rather than dropped: match … --strength monotonic bound the option, ran a read that never looks at it, and reported nothing — from the outside, indistinguishable from a strength that was applied. The three KB flags (--dir, --memory, --starter) go with any command, repl carries the union since its options are fixed at start and every line reuses them, and help names the owner of each. Refusals print on stderr, so a script reading stdout as EDN gets data and the terminal gets the diagnostic; the REPL loop keeps stdout on purpose, its errors belonging in the transcript. Class: Breaking for the stream move — a script capturing only stdout stops seeing refusals — and for the per-command roster, since a line carrying a flag its command ignored now exits 1; Refusal for the flag-as-value. Migration: redirect with 2>&1 if you were reading refusals off stdout; drop the flags your commands were ignoring; re-assert any rule whose --strength was dropped if you read the slot back.
  • test-backends.sh gives each run target/test-backends/run-<pid> with a latest symlink, two concurrent matrices having deleted each other's live disk scratch and interleaved one log, and its header names the :fan four-assertion divergence instead of calling every count difference a skip. The shellcheck roster gains test-parallel.sh and run-bench-caches.sh, the two drivers nothing checked.

Three more costs read the change rather than the KB. The pattern the except roster and the planner fan above are the largest cases of.

  • The overlay's removal record asks the base kv-member? instead of materializing the whole posting per probe, the rule merged-member? already states. refresh-equality walks the moved handles through a handle-to-edge reverse map — the moved-edges discipline the transitive relations hold — instead of re-asking belief of every supporter and re-deriving every edge per merge. And the qualitative join baselines live in their own bounded map, so the resident cache clearing at its limit no longer degrades every later delta join to a full one. Class: neither label; each answers what it answered.
  • settle's clash-candidates sorts the moved region only when something reads the order. It sorted twice per settle — content-order over revisit and over touched — before reading checks/arbitrating?, and the only reader of that order is the budgeted declaration sweep below it, which runs under arbitrating?. So a :refuse KB, the default, paid two sorts per settle to feed a pass that was never going to run.
  • core/check's docstring says what it predicts and what it does not: it answers assert's refusals under the KB's current constraint policy, and arbitrable is narrower than clashing. Guard: constraint_policy_test runs its cases under both policies.

0.5.1 — 2026-08-11

What a write pays, and what an instrument can see. A run of costs that grew with what the KB holds rather than with what the write touched — the taxonomy reconcile, the five flat caches, the reified-NAT orphan sweep, a retraction's teardown, the standing-clash ordering, a context-cycle repair, a repeated closure ask and a query plan's child count — each now reads forward off the region a settle moved, and each has a lein perf check standing where the claim is. Four places where arrival order decided an answer are closed: a predicate's second declared inverse, kb-quality's capped lists, the standing clash reports, and a preview's capped diff. Beside them the process gained instruments for the rest — the change feed crosses the process boundary as a subscription with a cursor, long work is a job registry with a screen that watches it, kb-quality reads the knowledge where every other instrument reads the engine, and the conjunctive planner costs a join rather than a column of literals.

No entry is Breaking, which is why this is a patch. Three carry a Migration line anyway, because a caller can observe them and should be told what to expect.

Triage, for a 0.5.0 caller. This is the index to what touches something you have written.

If your code…Then
reads the first N of preview / edit-with-consequences!'s :believed-added or :believed-removedyou get a different N — the halves are content-ordered now, and were handle-ordered
calls clear-caches and expects the literal cache's hit rate to zeropass {:counters? true}; the reset is off by default
walks a declared-transitive predicate that also declares an inversethe walk sees the inverse-recorded hops too, so an ask can answer more
branches on violations' :violation with a defaultless case:arbitration-truncated is a new kind
builds on the shipped Space or Time vocabularyan argument position that held thing now names a type, so an assert 0.5.0 accepted can meet an :arg-type refusal — widen the convicting declaration it names, or state the argument at a type the position admits

Belief, and what a settle pays.

  • A preview's capped diff is ordered by content, not by handle. preview and edit-with-consequences! built :believed-added / :believed-removed off a region sorted numerically — handles, so assertion order — and :max-results took a prefix of that, so the browser's 50-row panel showed a different fifty depending on the order the KB was loaded in, with :bounded? true either way. Both halves rank on sentence then context (diff-order), at the point each caller caps. Migration: a caller reading the first N of either half gets a different N — a different, better-defined N. Class: neither label; the sequence moved with load order, so there was nothing stable to hold as a contract. docs/preview.md.
  • The pass that decides told nobody when it ran out of budget. A declaration arriving after the content it convicts reaches back over that content, bounded by settle/*exposure-instance-budget*. The reporting half filed :exposure-truncated; the arbitrating half, which spends the same budget before anything is decided, said nothing — so a KB could leave standing a pair a finished sweep would have defeated and show a clean ledger. It files :arbitration-truncated (:triggers :sample :budget :message), one entry per settle. The pairs past a cut go undecided this settle rather than decided the other way, and discovery accumulates in :clashes, so a later settle's region can surface them. Class: Additive — both renderers read (name (:violation v)) rather than dispatching on a closed set. docs/taxonomy.md, "Neither cut is silent".
  • Every mutation sorted every standing clash report. record-clashes! ordered conflicts and contradictions by content on the settle path, so an assert into a KB holding N standing dilemmas paid O(N log N) to order a reading nobody had asked for. The vectors are stored in arrival order and settle/ranked orders at the read: 1.60 → 1.07 ms per assert at 800 standing dilemmas, against 1.05 ms before the ordering existed. perf's negation-arbitration reads 10.02x against its 12x bound, from 12.65x failing. Class: neither label — both readings answer what they did, in the same order.
  • A belief flip cost what the taxonomy held rather than what moved. refresh-relation read the region backward: deciding whether to run walked every supporter, and running recomputed every edge's believed-supporter set — 176.6 ms in a 64k-edge relation, and 6.87x across 8x the edges the flip is not about. :handle-edge, the transpose of :support, lets the scope be read forward off the moved handles: 9.2 µs at 64k edges, 0.61x across the same 8x. Gate: taxonomy-belief-flip. docs/taxonomy.md, "The belief reconcile is scoped to the moved region".
  • Every settle paid the size of the declared vocabulary to learn it had nothing to reconcile. :cache-support holds every disjoint pair, predicate property, inverse and declared arity at once, and the five flat caches read it backward: the gate miss — nearly every settle — measured 5.0 ms at 32k declarations for finding nothing, and a flip 95 ms. Read forward off :cache-handle-keys they are 5 µs and 1 µs, flat across 64x the declarations. Gate: flat-cache-belief-flip. docs/taxonomy.md, "The belief reconcile is scoped to the moved region".
  • What the scoping removed was a rescan four writers were leaning on. add-edge / del-edge and support-add / support-drop run with no believed? in hand, so they recompute contexts from every recorded supporter — the believed reading on a single supporter, a superset on a shared one, which the whole-KB rescan corrected unasked. A writer touching a shared edge or entry records it in :dirty / :cache-dirty and the reconcile takes those whether or not belief moved there: a set proportional to the edits, never to the KB. Without :cache-dirty the flat-cache oracle fails 95 assertions.
  • The index behind the flat caches is a multimap where the closures' is 1:1. A genl sentence names one edge, so :handle-edge can be {handle edge}; the flat-cache writers key one entry per sentence but nothing in the structure says so, and a 1:1 index would let the first support-drop take a handle out from under an entry the same sentex still supports.
  • The two caches that still gate read whichever side is smaller. The equality partition and the rewrite rules hold asserted term-identity claims rather than vocabulary, and the equality scan rebuilds relation-wide state, so both gate rather than scope. moved-touches? compares the two counts and walks the smaller — 0.9 µs at every size, from 4.9 ms to decide a 32k-entry cache was untouched.
  • A taxonomy edge retired every memo, because a counter says something moved without saying what. At 400 standing dilemmas one genl edge with nothing above or below it cost 800 checks/arbitrable-violations calls, and one genlContext edge under a context no contradiction was stated in re-derived all 400 opposed bodies. Both cost zero: clash-nogoods weighs the genl relation per pair, keyed [type context], and negation-nogoods records the joint-visibility verdict per context pair and re-derives only what moved. Gates: taxonomy-edge-arbitration and context-edge-arbitration.
  • Two more settle-path guards charged the whole KB. refresh-supersessions runs every settle and re-examined every displaced spelling each time — one probe per standing merge per write, which is what an owl:sameAs import emits. Narrowed to jtms/touched plus migration's own output, one unrelated retraction against 400 standing merges falls from 400 rewrite-term* calls and 9.06x to 0 calls and 1.70x. Beside it, a negated exception conjunct registers under the functor not, which hides the predicate it is about, so the genls(super) walk waved it through to :all — one level-6 query per firing the rule had ever made, per genl edge written anywhere. Keyed on specs(sub) instead: at 800 firings, 1,600 exception evaluations and 10.16x become 0 and 0.91x. docs/equality.md, docs/exceptions.md.
  • A disjointMetatype's membership is vocabulary, not a roster. clash-vocabulary carried the metatype roster and not the metatype membership: (disjointMetatype M) separates M's members by being consulted, so (M b_t) leaving stops separating a_t from b_t while the mark still stands, no closure moves, and neither member of the standing pair is in the region — so the KB kept reporting a dilemma the oracle does not. Only the departure was silent, a member arriving reaching its pairs through metatype-member-reach.

Where arrival order was deciding an answer.

  • A predicate's second declared inverse hid its first, and which one survived was the order they arrived in. (inverse P Q) and (inverse P R) are both legal and the taxonomy held one partner per predicate, so (transitive beforeEv) proved (beforeEv A C) with afterEv declared second and failed with it declared first; retracting one dropped the predicate's whole entry, leaving P with no inverse while (inverse P Q) was still believed. :inverse is {predicate -> #{partners}} now, maintained in both directions as :disjoint-index is. tax/inverses-of is the set the step relation and solve-inverted read; inverse-of keeps its shape — one partner or nil — and answers the lexicographically smallest. vaelii.core/inverse-of is unchanged in arity and return. Class: neither label — nothing documented which of two declarations won. docs/taxonomy.md, "The step relation".
  • kb-quality ranked its capped lists on the handle, which is assertion order, so two loads of the same knowledge reported the same :never-count over a different :never. The listed sets rank on content — consequent predicate, then sorted antecedent predicates, the stored sentence breaking a tie within a signature — so the store is read for the listed rules rather than for every rule. Class: neither label; kb-quality is new in this release.
  • A prompt is cut by content, and a sample says it is one. The LLM prompt builders took their lines and then sorted them, so a term with more mentions than the cap was shown a prefix of arrival order, and the same knowledge loaded twice proposed against two different pages. The sort precedes the cut at all four sites, and the heading tells the model it is looking at a sample. :max-scan (4000) replaces a hidden 4x oversample. used-with's own scan cap is the one place arrival order still shows, and closing it means walking the extent. Class: neither label — vaelii.impl.llm.* is impl. docs/llm.md.
  • The scan above the cut was ordered by the index, which is not an order at all. Those four cuts sat on a take over the term index's posting set — hash-ordered, moving with the representation, and differing between the columnar and KV backends over identical knowledge. The handles are sorted before the cap, which costs a sort and no extra record reads. Below the bound a page is a function of the knowledge alone; at the bound it is a sample of the term's earliest mentions, ranking by content meaning fetching all of them.
  • A card's cut is a count, including the cut that was not counted. used-with claimed everything its scan missed was "still offered under a later tier", which is false for exactly the predicates it exists to find — one used with the term but never declared and never argIsa'd lives in that tier alone. :dropped gains :unscanned, one O(1) count-with-arg per position, and the card says in words that it did not read N further facts. Beside it, correct.clj took first of a position's argIsa declarations, so two contexts declaring one position decided by index order whether a reversed-argument alternative was offered; it ranks by specificity now, narrowest first.
  • A declaration's supporters were being lost or arbitrarily chosen, and the cycle goal could throw. The disjointMetatype sweep walked the believed memberships where it records supporters, so a membership defeated at that moment never entered :cache-handle-keys and clearing the defeat could never revive it — and the live KB disagreed with the recovered one over the same store. It reads stored-declarations now. derive-functional-equalities took first of a (functional P) stated in two contexts, so retracting that one withdrew a merge the other still licensed; every supporter contributes a justification (tax/prop-supporters). And (P ?x ?x) ranked answers with a bare sort, which threw ClassCastException out of ask on two structural terms; it is sort-by pr-str.

The model backends, the apply path, and what they refuse.

  • A credential that cannot ride in an HTTP header is refused by name and never by value. The JDK rejects a CR, LF, control or non-ASCII header character with an IllegalArgumentException quoting the value verbatim, and the browser renders an error onto the proposal panel — so a .env ending in CRLF was one hop from putting an API key on a page. The header call is re-raised as {:type :llm-bad-credential :header "x-api-key"}, carrying no value, and the environment reads are trimmed so the common case never reaches the JDK. Class: Additive: a :type where none was.
  • A streamed body has a deadline, so a host that goes quiet releases the thread. A request's .timeout bounds the response arriving, and a streamed turn is almost entirely what comes after — the path the browser always takes. Both transports read the body under a daemon watchdog that closes it at :timeout-ms, measured from before the send, and a failure after it fires is {:type :llm-timeout :timeout-ms n}. Class: neither label.
  • A 200 whose body is not JSON carries a :type like every other refusal here. A proxy answering HTML, or a chunk truncated mid-object, escaped as the JSON library's own exception — the one thing here a caller could not discriminate on. Both transports raise {:type :llm-bad-response :status s :excerpt …}, bounded at 200 characters and in the data. Class: Additive. Beside it the ant credential probe is killed at five seconds rather than waited on, and a backend that probes available and then will not build logs a warning naming the backend instead of falling through to the stub in silence.
  • A turn the model ran out of tokens in was being diffed into proposed deletions. :stop-reason was read only through refused?, so every row the model never reached came back as a :remove — a transport artifact rendered to a human reviewer as an intended retraction. Truncation is decided before the diff: :truncated is a status on the diffed paths (carrying no batch, and ahead of the tool-use arm so a half-written tool call cannot run) and :answer-truncated? a flag on the additive ones. propose-page lifts :page-found and :page-truncated? into the returned map where a select-keys cannot drop them.
  • apply-proposal! catches, settles, and says what landed. It calls edit!, which is documented as not a transaction, so a throw at entry N left 1..N−1 stored and skipped the settle — a KB holding a prefix nothing had reconciled, reported as a bare exception. It returns {:result :applied :failed-at :error}, with the settle run by hand on the failure path.
  • A StackOverflowError reading model text read as "the model proposed nothing". Every read of model-written EDN caught Exception, and a deeply nested form throws an Error. Eleven sites catch Throwable now, and the depths were measured rather than assumed: the EDN reader overflows around 5,000 nestings and sentex/canon at 500, well inside what the reader accepts. One site marked its parse failure with (.getMessage e)nil for a StackOverflowError — so the nil fell through to the map branch and returned an empty batch: a crash presented as a model with nothing to say. It names the class now.
  • The ^:llm consent gate could be satisfied by a helper that does not consent. The meta-test keeping the mark and VAELII_LLM_LIVE in agreement accepted a call to any fn named live-model, and one namespace defined one that was a bare provider constructor with no gate in it. The scanner follows the call now: it collects, to a fixpoint, the names whose own source reaches live-llm?, and takes defspec as a head too. Six marked tests, all proving consent.

The taxonomy and its closures.

  • A genlContext edge out of a context that sees another one back never returned. The depth potential ranks the condensation, so A and B are level as one component, and raise-depth lifted the single node — which put A above its own mate, round the cycle without end. The lift moves whole components, and the condensation being a DAG terminates it. taxonomy_depth_test bounds the call on a daemon thread rather than trusting it.
  • Retracting one edge of a context cycle cost the whole context graph. A deletion can split a strongly connected component, so deactivate surrendered the depth potential and rebuilt every component: 11.19x across 16x a background the retraction is not about. An edge can only break the strong connectivity of a component it belongs to, so an edge merely incident on a cycle is left alone; the same retraction reads 0.97x. Gate: retract-context-cycle-scaling.
  • A repeated closure ask reads the answer instead of walking it again. TransitivePredicateProver's open-argument arms hold the reach per [direction predicate node context], stamped with the change clock: 0.10–0.14x on a repeat over a 2,000- to 8,000-node chain, and no record read at all. A closed goal reads the cache without filling it, computing a closure to store charging a two-hop question for the whole extent. The clock is the whole invalidation story and is what makes the cache follow belief. Additive: no answer moves. docs/taxonomy.md, "What is cached, what is not, and why".
  • (P ?x ?x) cost a closure per node to answer nothing. The one-variable arm asked reaches? of every source term, and reaches? walks that source's whole reach in order to fail — so an acyclic chain cost O(n²) to answer the empty set it always answers. One iterative Tarjan condensation over the step relation now (on-a-cycle, O(V+E)). The answers are unchanged.
  • A transitive predicate's hops are the believed matches, and the inverse spelling is one of them. (transitive before) walked stored before facts, so (inverse before after) left every after-recorded hop off the graph — the chain broke mid-walk and answered negative with no diagnostic. succs / preds-of probe the partner literal beside the direct one, as a matches-visible call rather than a goal handed back to the registry. (inverse P P) is a legal declaration and now a storable one. Migration: an ask over a declared-transitive predicate returns at least what it did and never less, so the only caller affected is one that counted on an inverse-recorded hop being invisible — including through unknown and exceptWhen, which read the same list. Class: neither label, on §3.8's own precedent. docs/taxonomy.md, "The step relation".
  • The closure walk's per-edge cost is measured, and the record fetch is the minority of it. lein bench-walk times an n-node walk against a direct sweep over the same records on the same mount. What it finds is a threshold rather than a slope, and the threshold is the hot-record LRU's capacity: under it a :disk fetch is 0.06 µs and cheaper than the :memory one; past it the fetch is a page-in at 3.03 µs and reaches 21% of the hop. docs/taxonomy.md, "What one hop costs", docs/storage.md.

Reified NATs, and what a retraction sweeps.

  • A unary fact about a reified NAT was deleted with the constant, silently. One clause of the orphan sweep matched on arity alone, so (prime (PrimeFn Seven)) — a claim somebody asserted — made K look orphaned once its other uses went, and the sweep retracted the claim with the constant. No error, no report. Bookkeeping is decided by authorship now: nat/minted-for re-derives what mint-nat! wrote from the same believed declarations, and everything else naming K is somebody's assertion whatever its arity. The one constant this keeps that arity would have collected is one whose resultIsa was retracted after the mint, and holding it is the direction to err in. Class: neither label — the surviving set moves in both directions, but a caller whose unary claim was being deleted underneath them was not getting what they believed.
  • One retraction cost what the whole KB had ever reified. The sweep asked "which constant is orphaned?" by matching (termOfUnit ?k ?e) — every NAT in the KB — after every teardown and to a fixpoint: 16.70x across 16x the NATs the retraction is not about, linear, which on a corpus of OpenCyc's order is seconds per retraction. No benchmark saw it, the sweep being gated on a declared reifiableFunction that no synthetic probe declares. It asks only about the constants the teardown's own removals named, one inverted-term-index read apiece: 1.22x, gated by retract-nat-scaling.
  • A teardown records only what a sweep will read. integrate/removal-sink retains every sentex that leaves the store for the length of a teardown — on a cascade, the whole cascade in a vector — and it exists for a sweep that is itself gated on a declared reifiable function. The gate decides whether the record is kept, so a KB that reifies nothing pays nothing. edit! reads the sink before its adds, so a batch declaring the KB's first reifiable function finds a nil sink and takes the whole-KB arm.
  • The removals reach that sweep through integrate/*removed-sink*, arriving from three places of which only one is the caller's: the dependency-directed sweep, the settle that follows, and the orphan sweep's own retractions — the third being what makes the region grow with the fixpoint, so a NAT nested in a collected orphan's expression is collected rather than left dangling. A use that merely stops being believed is not a use that went.

The change feed, and the daemon under load.

  • The change feed crosses the process boundary, as a subscription with a cursor. watch takes a function and a function does not cross an EDN wire, so the daemon's half is what request/response can carry: :watch answers a token, :poll reads that subscription's ring forward from an integer cursor, :unwatch drops it, :watchers says what is open. All four live in serve/feed-ops rather than serve/ops, which is what keeps a subscription out of the model's tool set. :lagged is on every reply rather than only the bad ones, and {:wait-ms n} parks the request outside the write monitor, capped at 30 s. docs/feed.md, "Across the wire".
  • A parked long poll held a thread nothing counted, so the feed could stall the daemon it exists to keep live. 55 concurrent polls drove ring's 50-thread pool to 50/50 busy and /health from 62 ms to 25,997 ms, one subscription being enough since nothing bounded polls per token. subscribe/max-parked (16) bounds how many may wait and serve/http-threads (50) is stated rather than defaulted so the pair is checkable. Over the ceiling a poll asking to wait is refused (:too-many-waiters, 400); one that does not ask, or whose events are already there, is never refused. Class: neither label — the feed is new in this release.
  • Two ways the feed could be taken down or lied to, and the bounds that stop them. {:wait-ms 1e300} answered 500: validated as number?, which admits a magnitude no long holds and admits ##NaN, which coerced to 0 and made the long poll answer instantly forever. It is nat-int?, capped before the coercion. And a subscription dropped while it was being registered took the whole feed down permanently: an unguarded assoc-in recreated an entry with no :polled-at, so reap — at the head of every feed op — threw for every later call on that daemon. What one caller can allocate is bounded three ways, nothing authenticating POST /op on the loopback default: 64 subscriptions, 256 events per ring, and one unpolled for five minutes reaped at the next call. A ceiling refuses the new subscription (:too-many-subscriptions) rather than evicting somebody else's. Class: neither label.
  • The ceilings bound the event count, not the bytes, and docs/feed.md says so: an event carries one settle's whole relabelled region, and 20 batches of 500 facts left one abandoned subscription holding 10,000 preview entries. :watchers reports :delivered and :pending — both client docstrings said cursor, the one word neither may use, neither number being the reader's position. vaelii.client/watchers, vaelii.serve/feed-ops and vaelii.serve/op-names join public_api_test's roster, a subset check that had been letting them through unnamed. The browser is untouched, and that is a decision: its live regions poll an htmx fragment, and a job's percentage is not belief moving.

Conjunctive query planning.

  • Planning one fixed conjunction is flat in the size of the KB, and a gate says so. The cost model divides by the trie's distinct-value count at a position and asked for it once per literal per plan, and (count (children …)) answers that by materializing the child set — 25x more against 32x the facts, per rule expansion, per node, per prove call. p/count-children is the same number off a cardinality. Additive — a new IndexStore read, owed by the two implementations of that protocol. plan-scaling reads flat with the count and 24.6x without; the one exception is the overlay's merge rule. docs/indexing.md, docs/overlay.md.
  • A conjunction is costed as a join rather than as a column of literals. plan/est-rows answers the question est-matches does not — not can this literal fan out, a sound upper bound, but how much, an expectation wrong in both directions and composing for exactly that reason. It returns the relation's shape ({:rows 400 :vars #{?x ?y} :distinct {?x 20}}) and the planner threads it through its fold, so the k-th pick is costed against the rows reaching it. On three or more antecedents a per-literal count's error is multiplicative. docs/inference.md, "Conjunctive query planning".
  • A count the trie read beats one inferred. :distinct holds only the column the walk can reach, and the join divides by the larger of the counts the two sides read; where neither read it the model falls back on a proxy rather than calling the join a cartesian product, which is the error that compounds fastest with depth. No statistics table: every number is already in the count-aware trie. The generators are split into blocks — two literals sharing a variable are one connected component — ranked by adjacent transposition, a descending sort on s/(n−1), with two placements outside the law because they are claims an estimate cannot make: a block proved to match at most once runs first, and so does the block reached by the caller's bindings, the evaluables or the recursive literal.
  • The estimator is measured before the plans are, and lein bench-plan reports the curve. q-error per join depth is the go/no-go: flat in the depth means the estimates compose. It reads 1.00 at every depth through six literals on a uniform chain, and 1.00 / 2.75 / 2.87 on a corpus built to break the independence assumption — wrong, and flat, where a compounding model would read about 7.5 at depth 3. Against an oracle over all 24 permutations of randomized four-way joins the planned order runs a mean 1.13x the best possible (2.18x before), and the same conjunction as a rule's antecedents runs 7.1x and 8.4x faster planned at four and five antecedents. Two assumptions are stated rather than pretended away.
  • query-plan carries the numbers the order turned on. Additive: :est-rows, :est-prefix and :block beside the :est-matches already there, in the browser's plan table too. A literal placed early on a small :est-matches whose :est-prefix then jumps is the model being wrong about a join, not about a literal. They are read off the plan that ran.

The browser: jobs, caches, and a reading of the knowledge.

  • Long work is one mechanism: a job registry, and the screen that watches it. vaelii.impl.jobs holds every operation that takes minutes rather than milliseconds, with one status vocabulary (:running:cancelling:done / :cancelled / :failed), one progress reading and one cancel; the browser gained /jobs, /jobs/rows and /jobs/cancel. A job survives the request that started it, a finished job's report stays an hour, and nothing unsettled is ever dropped, since forgetting a job releases its writer claim while a thread still running is still writing. The catalog's load and export were moved onto the registry rather than left beside it. docs/web.md, "Long work as jobs".
  • POST /chain is a job, with the derivation bound on the form. A fixpoint over a corpus was minutes of this process's one writer inside a request, with nothing on screen and no way to stop it. It reports about four times a second, takes a cancel at its next report, and carries :max-derivations; a run that settles inside 250 ms still answers with the /stats page, so nothing small acquires a spinner. What a stopped run leaves is stated beside the button.
  • A second writing job is refused as :job-busy, where a second load answered :busy. One job writes at a time — a load and a chaining run each claim the writer, an export claims nothing — and the refusal names the job that holds it. Not Breaking: :busy was thrown at one impl site and read by one impl namespace. It carries an entry because type_contract_test holds every :type in the sources.
  • The browser's status words are the registry's, so /kbs says running and done where it said loading and ready. A screen-scraper is the only caller that can observe it.
  • What this process is holding, on a page — caches, heap, and the profiler. Nothing measured what the engine holds beside the store, which is a dozen derived structures whose whole purpose is that a repeated question is not recomputed. caches is one read over all of them — entries, the bound they are cleared at, what one entry counts, and the hit rate where anything counts one — rendered at /caches. A hit rate is the cost model's report card. VAELII_PROFILER starts clj-async-profiler's UI through a requiring-resolve, so it exists without the dependency and says so when the class is absent. docs/web.md, "What this process is holding".
  • A number's scope is part of the answer, because two of them differ. A row carries :scope for what its entries count and :counters for what its counters count: the literal cache's entries are one KB's and its counters are global, so rendering the second as the first attributes another KB's work to this one. :unit is on every row for the neighbouring reason, and :limit takes a thunk where the bound is a dynamic var.
  • The list is complete rather than merely finite. Every cache-holding namespace declares itself into a register at load, so there is no central list to forget — and a cache in a namespace this process never loaded has no row, which is the honest answer where a row of zeroes would claim a cache that does not exist. A row that throws is reported as a cache that could not answer, and costs its own row and no other.
  • The clear is a measuring instrument, not an edit — and clear-caches took a KB and reached past it. It drops every derived cache and reports what went; nothing is destroyed, so it is bare rather than !, moves no belief, holds no writer, and is usable while a load runs. It also reset the literal cache's process-wide hit and miss counters, so asking one KB to drop its entries zeroed the rate every other KB in the JVM was reporting. That reset is {:counters? true} now, off by default, with :counters-reset in the reply. Class: neither label — clear-caches is new in this release.

The shipped ontology.

  • The shipped ontology decontextualizes predicate metadata and nothing else. SocietyContext declared (decontextualizedPredicate marriedTo), the one domain relation carrying a mark the rest of the ontology reserves for claims about a predicate — so a marriage stated anywhere became a claim of the whole KB, and a story, jurisdiction or hypothesis could not hold one the rest of the KB did not share. It reached past what the declaration names, too: a rule fires on the lifted copy, so SocialContext's marriage rule put knows within reach of every data context. The declaration is gone; marriedTo keeps symmetric, and a KB that wants marriages lifted asserts the declaration where its reader can see it. Class: no label — ontology content, not the surface a caller writes against (§3.8). docs/contexts.md, "What the shipped ontology declares it of".
  • Every argument position in the shipped ontology is declared. 227 argIsa / argGenl declarations across CoreContext and the four upper contexts that carried none, at thing throughout — the point being that every position is stated, not that any is narrowed. What it buys is schema completeness, a wrong-position refusal, and an edit rather than an addition when a position is later narrowed. argGenl genl 2 is the one position left undeclared, and CoreContext says why beside it. docs/argtypes.md.
  • Six types enter the upper ontology, and the declarations narrow onto them, so an argument constraint refuses something rather than only recording that the position was considered. spatial_thing takes physical_object beneath it, time_point sits under temporal_thing, function beside predicate, and integer, character_string and context name themselves. Space takes spatial_thing on all 100 of its positions, Time temporal_thing on 46 and time_point on 16, and the meta-vocabulary takes predicate, function, context, integer and character_string where it held thing — so (before genlContext genl) is an :arg-type refusal. Class: no label — ontology content (§3.8); what it owes instead is the roster that pins the shipped set, vocabulary_audit_test. Migration: a KB built on the shipped Space or Time vocabulary can be refused where 0.5.0 accepted it — the refusal names its convicting declaration, so widen that or state the argument at a type the position admits.
  • Flight is a capability of a kind, and the kind that cannot is an exception. flies was a verb-shaped one-place predicate, which says the thing in a shape that cannot be generalized. It is (hasCapability bird flying) now, and canTravel goes the same way — travelling is a capability and flying a kind of it, so what flies travels off the genl closure rather than off a second rule. Three nouns enter the upper ontology: capability under intangible, with flying and travelling under it. hasCapability is read at both levels and carries no relationKind: (hasCapability bird flying) says the kind flies, (hasCapability Tweety flying) says one bird does, and a rule joins them rather than either being the other. The exception is written twice, because there are two things to except: (not (hasCapability penguin flying)) at the kind, which leaves crow and eagle flying, and an exceptWhen on the descent rule at the member. Class: no label — ontology content (§3.8).
  • has-prop? and props accept every kind the engine marks. ::prop-kind named six of the ten kinds the special table records, so :asymmetric — which has-prop?'s own docstring and docs/api.md both list — along with :abducible, :reifiable and :unreifiable were documented calls that instrumentation refused. prop-entry records its kind in the table entry under :prop, so the roster is derivable from the vocabulary that defines it, and special-table-test holds the marked set and the specced set equal in both directions.

Storage, and the instruments that price it.

  • A bulk load is decomposed, and the index write is 57% of it. lein bench-loadphase loads one corpus repeatedly through the same door, each run with one more phase stubbed out from the outside in, so consecutive runs differ by one phase and the deltas sum to the baseline. At 1,000,000 distinct binary facts on the :memory pair — 43.4 µs/fact, 23,100 facts/s — the index write is 56.8%, the JTMS node and premise mark 21.5%, the special-predicate suite 10.5%, the public assert prelude 6.7%, the record store 3.5% and canonicalization 2.0%. Two decorators split the index write further: postings 35–39%, key streams 11–15%, count maintenance 6–10% — so the counts are priced and are not the lever. Two write-side tricks measured worse and are reported rather than built: a transient for the whole load ran 5–7% slower, and sorting by trie key buys locality a hash map has nowhere to spend. docs/storage.md, "What a bulk load costs".
  • The one write on that path that grows with the corpus is guarded, and the load reports its own rate. A store posts its sentence's body to the negation memo's :dirty set, one conj per fact into a set that ends a load holding an entry per fact. Both readers filter by :opposed first, so kb/note-opposed! writes only for a body opposed before the store or after it. It does not move the wall clock (0.994x at 250,000 facts); what it removes is a structure proportional to the corpus, which is a claim about a ten-million-fact load's heap. Beside it, bulk-assert-facts! reads :on-progress{:phase :loading :done n :elapsed-ms ms :facts-per-sec r} every 100,000 facts, and a closing {:phase :done :total n …}. docs/storage.md, docs/api.md.
  • The workload profile grows the two arms no reasoning workload runs, and prices a retraction. Every arm lein bench-profile had was a load, a chaining run, a proof or a synthetic probe, so the term index and roster read zero on every corpus — which reads as a family nobody uses and means a family no reasoner uses. The interactive arm makes the reads an application makes, and its read table is the inverse of every other arm's: 88% :term-index, 12% :term-roster on the shipped starter. The churn arm retracts a sample of premises and puts each back, the only way unindex-sentex! runs at all. A fifth tally :retracts catches what that costs, kept apart from :writes because :dead is decided by what else shares the prefix. On the starter a retraction is ≤23.8 batch ops against an assert's 18.1. docs/profile.md.
  • Eleven checks join the perf gate, and one of them was widened before it ever shipped. flat-cache-belief-flip, taxonomy-belief-flip, retract-context-cycle-scaling, retract-merge-scaling, retract-nat-scaling, closure-membership, plan-scaling, quality-report-scaling, genl-edge-negation-recheck, taxonomy-edge-arbitration and context-edge-arbitration — plus standing-clash-reading, which costs what the standing clash entry above moved. 27 checks in the vector, and all 27 judge. A read of the standing set is Ω(n log n) by construction, so its bound is calibrated from both ends: healthy it reads 85.4x and 80.9x over a floor near 66x, and with the read filtering by cross product — the shape the claim rules out — 937.8x. It ships at 175x. retract-merge-scaling reads 5.66x alone and 10.49x in place and ships at 18x, lein perf running one JVM over the whole vector: that position dependence is a property of the harness rather than of the engine.

0.5.0 — 2026-08-07

Operating the engine, in the two senses a running process needs: what it will let a caller do, and what it will tell an operator it is doing. The daemon authenticates and refuses to bind an address without a credential, ships as a container image, and says which posture it started in; every switch the build reads has a row in a table a test keeps honest, and the four that took a presence where they meant a value now refuse a value nothing reads; the log level is a dial a running process turns; and the failures that look like answers — a query that returns (), a KB that shares another's store, a sentence legal enough to store and wrong enough to never match — each gained something that says so. Beside all of that, two entries are about the vocabulary rather than the process: a name can carry two more roles than it could, which is what a KB built by reading text needs of it, and a reader that counted a context's contents was the only one of three that did not say count. Nine entries are marked Breaking, one of which simplifies a store's name from two numbers to one and one of which turns a switch's name the right way up. The three Refusal entries (CONTRIBUTING §3.8) cover input that is newly refused where what 0.4.0 did with it was run a configuration nobody asked for and report a clean pass, so no working caller loses anything it had. Each entry says what a reader would have observed; the mechanism is in the subsystem's doc, and the entry names it.

Triage, for a 0.4.0 caller. Every Breaking and Refusal entry below carries its own one-line Migration; this is the index to the ones that touch something you have written or deployed.

If your code…Then
names :record-space / :index-spaceone :space — keep the record number, drop the index one
runs a daemon on a non-loopback --listenexport VAELII_API_TOKEN there and in every client, or it exits 2
reads a daemon 401, or branches on the wire :type:unauthorized is a new one; GET /health is the only route without the token
relies on VAELII_RETE=0 running the sweepit means off now; unset, or =1 for on
sets VAELII_NOHIERit is VAELII_HIER, the other way up — VAELII_NOHIER=1 becomes VAELII_HIER=0
sets VAELII_QUERY_ENGINE / VAELII_QUERY_STRATEGYa name outside the roster is refused rather than silently running the default
sets VAELII_WEB_PORT for lein browser while -main stayed on 3000it moves both; pass --port 3000 to pin -main
lists KBs out of a search-path directory holding more than 200name the ones that matter in the catalog file
depends on vaelii and has no SLF4J provider of its ownadd one — org.slf4j/slf4j-nop no longer arrives transitively
branches on what term-role answers:sense and :lexeme are two new answers — add arms, or a default
writes a lex-namespaced predicateit names a lexeme now, and a lexeme names no relation
calls core/context-size, or sends the daemon :context-sizeboth are count-in-context — same arguments, same answer
compares two compound terms with differenta merged symbol inside one now makes them equal, where it did not
sets VAELII_ASP_SOLVER to a name outside clingo/claspit is refused at open-kb rather than silently running auto
  • Breaking: a name can carry two more roles — a sense, and a lexeme. term-role answers :sense for a disambiguated type (abrasive-grit) and :lexeme for a symbol in the lex namespace (lex/fool's_gold), so its documented domain gains two values a total case has no arm for. Two things move at a :strict front door with them: a lowercase dashed name is a legal unary type where it was refused, and a lexeme applied to arguments is refused (:lexeme-functor), a surface form naming no relation. Migration: a case over term-role gains :sense and :lexeme arms, or a default; nothing else changes unless you wrote a lex-namespaced predicate, which names a lexeme now and cannot be applied to anything. docs/naming.md.
  • Breaking: context-size is count-in-context. The three O(1) cardinality readers are one family and two of them said so; the third delegates to a protocol method already called count-in-context, so the name it now carries is the one it always answered to a layer down. The daemon's op keyword moves with it. Migration: (v/context-size kb ctx) becomes (v/count-in-context kb ctx) and {:op :context-size} becomes {:op :count-in-context} — same arguments, same answer, and the old spellings are gone rather than deprecated. docs/api.md, docs/indexing.md.
  • Breaking: different descends into compound arguments. It normalized each argument with one lookup in the equality closure, and the closure is keyed by symbol — so a compound was never found in it and (different (QuantityFn 5 Kilogram) (QuantityFn 5 Kg)) answered different with (sameAs Kilogram Kg) believed. It replaces symbols at every depth before comparing, the congruence its documentation always described. Migration: a goal comparing two compounds can newly answer false where a merge reaches inside one; comparing symbols is unchanged. docs/equality.md.
  • Breaking: one space number names a KB's stores, :space. open-kb takes a single number where it took :record-space and :index-space, and it defaults to 0; a :disk KB's derived directory is space-<n>, and the suite owns a block of two db numbers rather than four (scratch 15, isolated 14). Migration: {:record-space 2 :index-space 3} becomes {:space 2} — keep the record number, drop the index one; either retired key is refused by name (:type :unknown-option) rather than ignored. Pass :dir to name a durable directory the derived spelling does not reach. docs/storage.md.
  • Breaking: the daemon authenticates, and refuses to bind an address without a token. With VAELII_API_TOKEN set, every request carries Authorization: Bearer <token> or is answered 401 with {:ok false :type :unauthorized} — a new :type on the wire — and GET /health is the only route that answers without it. What the daemon binds decides what it requires: --listen naming a non-loopback address without a token is one line on stderr and exit 2, where 0.4.0 logged a warning and served the whole write block to anything that could reach the port. Migration: export VAELII_API_TOKEN for a daemon that names an address and give the same value to every client that reaches it; vaelii.client reads the same variable and takes :token. Nothing changes on the loopback default. docs/operations.md, "Daemon — vaelii.serve".
  • Every switch the build reads has a row, and a test keeps the roster honest. docs/operations.md gains a configuration table — 56 environment variables and JVM system properties, grouped by who sets one, each with where it is read, its legal values, its default, and the one thing it decides. config_surface_test pins the names against test/golden/config-surface.edn in both directions and checks each file:line citation against the line it names, so the table cannot drift from the code without a failing test. CONTRIBUTING §3.8 files a renamed or removed switch as Breaking.
  • Refusal: the four harness switches read a value instead of a presence. VAELII_RETE and the hierarchical-retrieval switch were membership tests, so =0 ran the sweep it names and an exported-but-empty variable ran one nobody asked for; the two query switches took a bare (keyword …), so a misspelt engine ran the default and reported a clean pass for a configuration nothing exercised — the worst shape a test switch can have, since the result reads as evidence. All four take the engine's boolean vocabulary now and refuse anything else by name, and a test calls each reader with the properties cleared so the table's Default column fails rather than merely reading wrong. All four are read only by the test harness, which is what keeps a silent change of sense inside a Refusal rather than making it a ninth Breaking. Migration: none for a value in the vocabulary; a job relying on =0 meaning on now gets the sweep off. docs/operations.md, "Developer — the suite and the scripts".
  • Refusal: the ASP backend switches are read against their domains, at the door. VAELII_ASP_SOLVER took a bare (keyword …), so a misspelt backend matched no arm of the selector and ran auto — a run pinned to clasp could use clingo and report a clean pass for a backend nothing exercised. VAELII_CLINGO_MAX_BYTES parsed with a bare Long/parseLong inside a cached delay, so a non-numeric value threw from the first ASP solve rather than from the configuration that was wrong. Both go through config/check! now — refused at open-kb, by name. Migration: none for a legal value. docs/operations.md.
  • Breaking: VAELII_NOHIER is VAELII_HIER, and the sense is the other way up. A switch that carries the negation in its own name makes =0 mean on, which is the one thing a reader must not have to work out at a glance — and the entry above had just made the value load-bearing, so the two had to move together or =0 would read as the fallback path it now selects. VAELII_HIER defaults true (the set-algebra retrieval), and VAELII_HIER=0 routes every context-scoped match through the reference nested fan-out. Migration: VAELII_NOHIER=1 becomes VAELII_HIER=0; a VAELII_NOHIER left set is simply unread, since a variable cannot be refused by name. docs/operations.md.
  • The log level is a dial a running process turns. vaelii.core/set-log-level takes one of :error :warn :info :debug :trace and installs Trove's console backend at it; log-level reads back what is in force, and VAELII_LOG_LEVEL says it at startup (a value outside the five is refused by name). Unset, the engine installs no backend at all, so an application holding its own taoensso.trove/*log-fn* keeps it. Three :debug statements are what make turning it up worth doing: what a chaining run concluded and how long it took, what a settle cost and found, and the rule a dropped conclusion came from. docs/operations.md.
  • Breaking: VAELII_WEB_PORT moves -main's port, and not only lein browser's. dev-repl read the variable and -main did not, so VAELII_WEB_PORT=3011 lein run -m vaelii.web bound 3000 and logged 3000. Both read one default-port now: the variable, else the vaelii.web.port property, else 3000; an explicit --port still wins. Migration: a deployment that set the variable for lein browser while relying on -main ignoring it now moves both; pass --port 3000 to pin -main.
  • Breaking: a search-path directory is probed for its first 200 entries, and a KB below the cut no longer appears on /kbs. sources is recomputed per request, which is what lets a corpus appear with no restart and what made the scan unbounded — a classify per candidate, and a size estimate per :store one, on every page load. catalog/max-discovered bounds it, and the cut is named on the page and in the log, since a list that quietly ends early reads as "this machine has no other KBs". Migration: name the ones that matter in the catalog file to list them regardless of the count. docs/catalog.md.
  • The front door says what a legal-but-wrong sentence should have been. (isa Muffet Dog) breaks no naming invariant, so it stored a two-place relation nothing reads and (isa? kb 'Muffet 'Dog) answered false with nothing to search for. nm/advice reads intent where problems reads the invariants: it recognizes the shape and logs a :warn once per process spelling the rewrite that was meant. Beside it, a :no-placement drop names genlContext and points at the :rule-context / :fact-contexts already on the entry. docs/naming.md.
  • A second open-kb defaulting onto the shared in-RAM space now warns, naming both fixes — give the KB its own number, or name {:space 0} explicitly to say the sharing is meant. A warning rather than a refusal, since sharing the space is how recover sees the same records and how a base is mounted. docs/storage.md.
  • Refusal: the CLI checks each command's argument count before it dispatches, and help names what each one takes. dispatch reached into args with nth, so lein cli assert '(dog Rex)' answered error: IndexOutOfBoundsException — true about a vector, no help to someone who left off a context — and a long line was worse, since the extra operand was dropped in silence. One table now carries every command's arity, operands and gloss, so check-arity! and the usage text cannot go out of step. Migration: none for a call already at the right arity; lein cli help prints the count each command takes. docs/operations.md.
  • docs/troubleshooting.md is a new page, indexed by symptom rather than by subsystem. The engine's hardest failures are the ones where nothing goes wrong — a query answers (), an assert returns a handle, and both are legitimate values no error distinguishes from the answer that was wanted — so a reader has to already know the cause to find the page explaining it. Nine symptoms, each with what you would have observed, how to confirm it in one call, and the fix.
  • lein lint gains a versions check, and the kondo row notes a local/CI version mismatch. The :with-foreign pin and defproject's own version are cut together and nothing held them to it: the 0.4.0 bump left the pin naming 0.3.0, so every lein with-profile +with-foreign command failed to resolve. lint-versions reads that pair and the lein-cloverage version stated twice, failing when either disagrees. The kondo row prints a NOTE — never a failure — when the local binary is not the version CI pins, since a newer kondo infers more than an older one flags.
  • Three doc samples now print what they actually produce, and prove's docstring says it counts proofs, not answers. prove returns one solution per derivation, so a goal reachable both as a materialized fact and as the rule concluding it comes back twice with equal maps — wrap it in distinct for an answer set, or reach for query / ask, which project to the goal's variables and answer each binding once.
  • The daemon ships as a container image, with a two-stage Dockerfile and docker-compose.yml: a build stage that runs lein uberjar, and a runtime stage of a JRE and the jar alone. The container binds an address, so the token is required — an image run without VAELII_API_TOKEN does not start rather than serving unauthenticated — and one container per volume, a second opener being refused :disk-locked rather than scaled, which is why the compose file carries no replicas:. docs/operations.md, "Container — the daemon as an image".
  • A reflection warning and an uncalled public var now stop the build. Both signals were already emitted and neither was read. lein lint gains two rows: reflect compiles src and bench and fails on any reflection, auto-boxing or primitive-recur warning, and unused reads clj-kondo's analysis over src test bench and fails on a public definition with no usage, against scripts/unused-publics-baseline.txt. Ten warnings had to go first, none in src. CONTRIBUTING §1.1.
  • Breaking: org.slf4j/slf4j-nop no longer reaches a consumer's classpath. It sat in top-level :dependencies, so every application depending on vaelii inherited it too, where it could win SLF4J's provider race against that application's own backend and silence it — the one thing a library must not do on a consumer's behalf. It lives in the :dev and :uberjar profiles now, so every entry point this repo ships still carries it while lein deploy publishes it as a test-scope declaration a consumer does not resolve. Migration: an application that ships Jetty, had no provider of its own, and relied on vaelii's to keep it quiet now sees SLF4J's "no providers" line again — add org.slf4j/slf4j-nop, or any other provider, as its own dependency. docs/operations.md.
  • A public --listen bind with no VAELII_ALLOWED_HOSTS now warns. Naming an address drops the Host allowlist to every Host answered — a deliberate default, since a reverse proxy legitimately sets its own and an operator cannot always enumerate it in advance — but nothing said so at startup. host-posture names the policy (:allowlisted / :open) beside the token question, and a public bind left unset gets its own warning, apart from the token and TLS lines so a reader knows which check is missing.
  • docs/troubleshooting.md and docs/storage.md now name :type :unknown-backend. open-kb throws it from five call sites — an unknown :backend sugar name, the one {:records :memory :index :disk} pairing the axes refuse, and an unknown :records / :index / :tms kind — and none carried a line in either doc. The new entry reads the other key each throw's ex-data carries to say which of the five it is.

0.4.0 — 2026-08-05

Correctness fixes found by reading the engine against its own stated invariants, in the places 0.2.0 and 0.3.0 did not reach: a backward-chaining loop guard that made a conjunctive query answer nothing, doors that disagreed about what they would accept, an index trusted without being checked against the records it describes, slots and keys that let arrival order decide belief, and derived caches a settle read one revival out of date. Thirteen entries are marked Breaking — they refuse input 0.3.0 accepted, rename what it exported, or change an observable contract — which is why this is 0.4.0. The Refusal entries (CONTRIBUTING §3.8) cover input that is newly refused where what 0.3.0 did with it was corrupt state or answer a different question in silence, so no working caller loses anything it had. Each entry says what a reader would have observed; the mechanism is in the subsystem's doc.

Triage, for a 0.3.0 caller. Every Breaking and Refusal entry below carries its own one-line Migration; this is the index to the ones that touch source you have written, so the rest can be read at leisure.

If your code…Then
hands assert text it did not read as EDNit is refused (:shape) — fix the producer
writes exceptWhen literals like (lives_in ?x cold_place)spell them to the invariants; re-check any rule 0.3.0 left bare
spells an edit! batch {:adds …}spell it {:add […] :remove […]} — the old key wrote nothing
names one of :record-space / :index-spacename both, or neither, in every opts map
passes :direction to assert on a non-ruleit is refused; a rule takes it and now acts on it
states one rule two ways (bare implies after a set/*Rule)the slots join by content; retract! and re-assert to narrow one
calls edit or edit-with-consequencesthey are edit! and edit-with-consequences! — the wire op stays :edit
matches :bad-opt, or a :shape from a non-map optsmatch :unknown-option
reads a dump's meta.edn dialectit is :vaelii
stores skolem witness names across runsthe names moved; rebuild from the assertions (export! / import!) rather than carrying both spellings
parses a daemon 500 for a client mistakeit is a 400 with a :type
writes (ist Ctx S) with other than three elementsit is refused with :shape
  • A conjunctive query could answer nothing while each of its conjuncts answered. [(anc Tom ?y) (anc Tom ?z)] was empty where (anc Tom ?y) answered twice, because the per-path loop guard grew for a whole frame and a queued conjunct is a sibling of the expansion, not a descendant. Silent in every direction: forward chaining and the node engine both answered, provable? said false, prove-within reported :status :complete, and the planner became semantic. docs/inference.md, "The loop guard's scope is the subtree, not the frame".
  • Breaking: assert refuses a sentence that is not an s-expression. A string — what a failed EDN read hands back, from impl.cli's read-arg and the daemon's :args — was stored, indexed and believed as an object no query can match; nil likewise; a symbol, number or map threw a bare UnsupportedOperationException with no :type. check refused all five, so the door built to predict assert disagreed with it. Migration: nothing a working caller sent is refused; fix the producer that handed assert unread text, and discriminate on :shape.
  • Breaking: an exceptWhen query's literals are held to the naming invariants. (exceptWhen (lives_in ?x cold_place) …) stored a literal docs/naming.md says is refused, as an exception no query could match — so the rule read as guarded and fired as bare. Both doors now read each conjunct, before the rule is stored, so a refused exception leaves no bare rule believed. Migration: spell the exception's literals to the invariants (livesIn, not lives_in), and re-check any rule 0.3.0 left bare.
  • Breaking: an edit! batch key nothing reads is refused. {:adds […]} bound nil, so edit! wrote nothing and reported {:added [] :removed {…0}} — a success — while check-edit, whose job is to predict exactly that, reported no problem. Over the daemon it was a 200 {:ok true} for a write that did not happen. Migration: spell the batch {:add […] :remove […]}; a batch under any other key wrote nothing.
  • Breaking: naming one in-RAM space number and not the other is refused. :record-space and :index-space default independently, so {:backend :memory :record-space 77} paired a private record store with the process-default index every other in-memory KB writes. assert then found the other KB's handle, read it as a duplicate, stored nothing, and returned a handle in? answered true for. A fork's :base and :overlay halves take the same keys and are refused the same way. Migration: name both or neither, in every opts map.
  • A durable index is checked against the records it claims to describe. layout.edn gates the index's key shape; nothing gated its coverage, so a short index opened clean, answered short forever and re-cemented its own stamp — and re-asserting a fact it could not find minted a second handle for a sentence already stored. Three ways in: a torn kv.log tail, a directory grown under a derived-index mode, and a crash between the record write and the index batch. docs/storage.md.
  • Breaking: assert acts on :direction instead of accepting and dropping it. Only assert-rule read the key, so a rule asserted {:direction :backward} stored :both and forward-chained, materializing the cross product a backward-only rule exists to avoid. A :direction on a non-rule, one contradicting the sentence's own wrapper, and a value outside the roster are refused rather than resolved. In the same pass a non-map opts answers :unknown-option from both doors, where check said :shape. Migration: spell the direction :backward (:forward :backward :inert :both); a check caller matching :shape for a non-map opts matches :unknown-option now.
  • Breaking: a re-asserted rule's direction and defeasibility resolve by content. Neither slot is in the identity key, so a rule stated two ways resolves to one record and the second spelling was dropped — letting arrival order decide a slot that decides belief. A bare implies after a set/inertRule stayed inert and never fired; after a set/defaultRule it stayed defeasible and lost to a monotonic rival it should have tied with. The resolution reaches conclusions already derived, since a justification bakes the rule's contribution in as its :strength at fire time. docs/canonicalization.md. Migration: the join only widens a slot; to narrow one, retract! the handle and re-assert the intended spelling.
  • The derived caches are reconciled with what clear-defeats! revived. A settle lifts last settle's defeats at its top, but the cached closures were refreshed only in settle-finish — after constraint-nogoods had read them — so discovery asked its question against a vocabulary one settle out of date. A P/¬P pair made visible by a revived genlContext edge went unarbitrated and retract! returned with both believed, a state recover over the same records disagrees with.
  • The disk KV index reads and publishes its RAM map under the lock. apply-ops! read @data before acquiring and published after releasing, while compact! runs on the durability daemon's executor — a thread the single-writer contract says nothing about — so a compaction in either window rewrote the log from a map missing the in-flight write. kv-clear! was sharper: a compaction between its truncate and its publish wrote the entire pre-clear map back over the log just emptied.
  • Breaking: a client's mistake answers 400 with a :type, not 500 with none. docs/operations.md promises every {:ok false} carries the type the engine threw; an unreadable body, a wrong argument count and an unknown op all answered untyped, the first two as 500s. The engine's whole refusal vocabulary now answers 400, unlogged — answered 500 they count as backend faults at every reverse proxy and 5xx alarm. Migration: a client branching on the status code should branch on :type; every {:ok false} carries a non-nil keyword.
  • The browser's /propose/* EDN read catches Throwable, as every other untrusted-EDN read in the namespace already does. A deeply nested form raises StackOverflowError, which an Exception catch let escape — and the browser has no exception middleware, so it left the handler entirely.
  • Refusal: query refuses a non-map opts and a negative or non-integer :max-depth. Both read as "no depth", which is not an error condition but a different question — the no-rule-expansion answer, returned as if it were the bounded one asked for. {:max-depth 0} is admitted: it is that answer asked for by name. Migration: none for a working caller.
  • Breaking: edit! refuses what check-edit reports, before applying anything. The two disagreed in both directions: a 4-element :add entry applied with the extra silently dropped where the dry run reported :shape, and a non-sequential entry threw a bare ISeq error from every door. An unknown :remove handle is refused before any entry is applied, so a checked-clean batch cannot half-apply. Migration: a remove-if-present batch filters its handles through in? first.
  • The recursive-literal hold-back keys on the peeled predicate. A not- or ist-headed consequent read its own frame as the predicate, so every frame-headed antecedent was "the recursive literal" — two orderings of a negated-head rule minted two handles, and a genuinely recursive rule with a negated head lost the hold-back, turning right-recursion left-recursive.
  • Breaking: a skolem witness is a function of its rule's content, not its handle. Retracting and re-asserting the same rule re-fired to a different witness, so a fact stated about the old one silently stopped co-referring — and two KBs holding the same knowledge in different orders stored different termOfUnit content, a handle in stored content that order independence rules out. docs/skolem.md. Migration: rebuild the KB from its assertions (export! / import! replays firings) rather than carrying both spellings.
  • Breaking: edit is edit!, and edit-with-consequences is edit-with-consequences!. The batch's :remove half runs the same retract-storage! sweep retract! runs, while the name read as additive — the one gap in the ! roster the convention exists to close. Migration: rename the calls; the wire op stays :edit, as :retract stays for retract!.
  • Breaking: :bad-opt is retired, and one compression spelling survives. Two keywords split one failure class on no rule a reader could predict — seven sites said :bad-opt where thirty-four said :unknown-option. Migration: discriminate on :unknown-option and :unsupported-compression.
  • Breaking: the dump's meta.edn names its dialect :vaelii. Decorative on the read side — the frame decides how a sentence is reconstructed — but it is a value in the frozen format and a documented key of import-dump's return, so the name it carries is now-or-never. Migration: a reader matching the old value matches :vaelii; import-dump reads dumps written either way.
  • The node engine's claimed-key reads each guard's identity, not the guard count. Two distinct rules, each carrying its own exceptWhen, can rewrite one goal to the same canonical residual through the genl fan; keyed on the count the two children were one key, so the second was dropped before it was enqueued and every answer only its exception admits was lost — silently, on the path query routes to whenever :max-depth is given. docs/inference.md.
  • A belief flip on a visibility except queues the same re-check as its arrival. Only the store and removal chokepoints called recheck-except, so an except defeated by a settle's resolution revived nothing it hid: backward proving answered yes while the store held nothing, and which belief set the KB ended with depended on the order the except and its defeater arrived.
  • recover reads only positive, atomic declarations into the taxonomy. sentexes-with-functor returns both polarities and the rebuild arms destructure the positive shape positionally, so a stored (not (genl a b)) bound its inner sentence as a taxonomy node and nil as the other — poisoning every cache on any recover, the default {:recover? :auto} reopen included.
  • A :neg nogood is an at-least-one in every reader. The ASP translation's soft branch emitted only the positive body atoms, so a :neg-only nogood — what set/softConstraint over negated choice literals produces — emitted its violation witness as an unconditional fact: no steering pressure, and :violated reported a satisfied at-least-one as broken. docs/solving.md.
  • conflicts and contradictions are content-ordered. Each report's sides were already ordered by content; the list came off a hash set of handle-keyed nogoods, so which pair (first (contradictions kb)) returned was an answer about which was typed first. docs/nmtms.md.
  • Refusal: the connective frames are shape-checked at every door. An implies at arity 2 threw a bare IndexOutOfBoundsException while arity 4 stored a silently truncated rule check read as clean; (not A B) stored as a positive fact whose record and index disagreed; a bare symbol passed as a rule literal was accepted, unmatchable; and a non-finite measure magnitude stored cleanly, then threw out of every later duration goal in the context. Migration: nothing a working caller sent is refused — every one of these stored an object no query could match.
  • Refusal: the last open rosters close. find-terms and abduce take key rosters (a misspelt :mtch ran the prefix default; a misspelt :keep? tore down the scratch context whose handles the caller meant to commit), the CLI refuses a flag outside its roster, escalate refuses a floor outside 0–7, and import-dump refuses an unknown :framing where it guessed a reader and failed as a ZipException. Migration: spell the key or flag as the refusal's roster lists it.
  • Refusal: the web and serve entry points refuse what their grammars do not know. vaelii.web --listen with no address parsed to a nil host — Jetty's wildcard bind, with the Host allowlist reading nil as any — so a truncated command line put the browser's unauthenticated write routes on every interface with the rebinding guard off. serve read its positionals as a prefix, so 4200 --listen 0.0.0.0 /var/lib dropped the directory and ran a disk daemon in memory. Migration: none beyond completing the command line.
  • Refusal: the opts and shape rosters reach the remaining doors. The roster guard held at assert, why, query and open-kb, and every other door took the misspelt key in silence — answering a different question than the one asked. Now refused: an open-kb mount or durability key without its axis, an opts key nothing reads at forward-chain, the extent readers, preview, export!, import! and the anytime budget maps. Migration: spell the key as the refusal's roster lists it.
  • Refusal: the operator's mistakes answer in one line. A CLI flag missing its value bound nil in silence — lein cli assert '(dog Muffet)' Ctx --strength stored known-true content at :default — and now exits 1 naming the flag; --memory --dir is refused as a contradiction. Migration: none beyond completing the command line.
  • The browser and CLI survive what they read. The repl loop and the CLI command arm catch Throwable, so a deeply nested form answers error: and a next prompt; the browser's retract POST makes the check-edit round-trip docs/operations.md promises, so a stale handle answers the problem panel rather than a success-styled "Retracted 0 sentexes".
  • Refusal: every durability switch is read against a domain, and a value outside it fails the open. Each of the thirteen checkable vaelii.* / VAELII_* switches was a membership test or an equality against one spelling, so none of them had a wrong value — every misspelling was the other branch, silently: vaelii.disk.auto-compact=disabled read as compaction on, and vaelii.disk.fsync=always as the three-second tick, the level the operator was trying to leave. The three numeric reads had no catch at all. Migration: none for a working setup, but two spellings now act where they were ignored — vaelii.disk.tokens=1 and vaelii.index.snapshot=1 turn their features on, and vaelii.disk.lock=0 disables the lock. Spell what you mean. docs/storage.md.
  • Refusal: the mapped index image refuses the platform it corrupts on. The image publishes by renaming a new file over the live one while it is mapped, which is what put vaelii.index.snapshot on macOS and Linux only — docs/storage.md said so and nothing enforced it, so on Windows the publish failed part-way through a four-file commit, naming neither the cause nor the fix. Migration: none — the property never worked where it is now refused. docs/storage.md.
  • Breaking: assert-rule refuses a rule literal whose predicate is a variable. (implies (and (?p ?x ?y) (transitive ?p)) (?p ?y ?x)) asserted cleanly and was indexed under ?var0, which no arriving fact and no goal can spell — so the rule answered no backward goal at all and fired forward only when the concrete-predicate antecedent beside it arrived. Two arrival orders, two answers, from a rule the engine reported as accepted. An :inert rule is exempt, which is what CoreContext's decontextualized- predicate lift is. Migration: assert the instantiated rules, one per predicate the metarule ranged over.

0.3.0 — 2026-08-04

Correctness fixes across the durable index, the snapshot, the JTMS, the export dump and the bounded prover, a sweep that gives every refusal a :type, the one wire contract 0.2.0's own sweep left qualified, and the serialization both servers' storage layer already assumed. Then a run of inference and belief work: two orders that reached two answers, the two doors that disagreed about an inherited claim, and two enumerations that grew with the vocabulary rather than with their own answer. Eight entries are marked Breaking — they refuse input 0.2.0 accepted or change an observable contract, which is why this is 0.3.0 and not 0.2.1; the rest are compatible.

  • Breaking: the daemon's refusal :type keywords are plain:not-edn, :cross-origin, :bad-host, :body-too-large, where the namespace serving them qualified each one. This finishes tree-wide what 0.2.0's own breaking entry claimed.
  • Breaking: both servers hold one request-body ceiling. The cap and its VAELII_MAX_BODY_BYTES override (16 MiB) live in vaelii.impl.guard, which both read, so the browser answers 413 for an oversized form body where only the daemon did. A daemon read is also fully realized inside the write monitor — wire-safe's walk is what realizes a lazy answer, so running it after the monitor released let a :query straddle a concurrent :assert.
  • Breaking: the browser serializes its writes. Jetty serves the write routes on a thread pool, so two POSTs were two writers — where the storage layer is written on the promise that they are not. Interleave two and the WAL holds both frames while the RAM map holds one, so the running index and the one replayed on the next open disagree. The browser now takes one process-wide monitor around every content write, as the daemon always did; a concurrent write waits rather than racing.
  • Every ex-info the engine throws carries a :type. Twenty refusals threw an untyped map, so a caller had to guess from which keys were present. Two forms that threw a raw Java exception now answer instead: (genl ?x ?x) / (disjoint ?x ?x) answer the question one variable in both positions asks.
  • Breaking: an ist form must have exactly three elements. 0.2.0 read assert and check positionally, so (ist Ctx S junk) asserted with the extra silently ignored and (ist Ctx) raised a raw IndexOutOfBoundsException. Both refuse with :shape.
  • The durable index is gated on its key layout at open. A log whose stamp does not match kv/index-layout-version is cleared, rebuilt from the records and restamped, :recover? notwithstanding; without the gate such a log replays cleanly and then misses every read whose key shape moved. A 0.2.0 durable store carries no stamp, so its first open under 0.3.0 pays one automatic reindex: O(records), logged at :warn, paid once. docs/storage.md.
  • Breaking: open-kb refuses a :base whose durable index is at an older key layout (:stale-index-layout). The repair is a write and a base is mounted read-only, so the refusal names the one place the rebuild can happen: open that directory as a KB, then mount the fork over it.
  • Breaking: (fork (fork base)) is refused (:stacked-fork), which is what docs/overlay.md has always stated.
  • Breaking: open-kb refuses a :recover? setting it does not name. :auto is the default, true an alias for it, :warn and false the rest; any other value read as the warn branch and handed back an empty TMS over a store that is not empty, which answers [] to everything. A stale derived index is dropped on open whatever :recover? says.
  • Breaking: close! releases a durable fork's own directory. A fork's writable half takes the same exclusive lock as any durable KB, so without its own :dir it could never be handed to another process short of exiting the JVM. 0.2.0's docstring promised the opposite, so code that closed a fork in a finally and kept reading it worked and now does not.
  • A failed compaction takes its temporary files with it. A rewrite that threw closed its handles and left <log>.compact behind, and the next compaction in the same session opened that temp and appended to it — its replay then put back records deleted in between. The cleanup is scoped to the pre-commit phase: past the marker the temps are the only complete copy.
  • A failed open gives back the directory lock with no handles still on it. open-kv-backend and open-token-log replayed their logs outside any guard, so a torn frame propagated to a caller that answers a failed open by releasing the lock — leaving it released while this JVM still held an open handle.
  • A fork's merged kv-entries is realized under its monitor. Both halves were lazy, so the seq handed back from inside the lock realized outside it. An export of a fork taken while anything wrote it projected two states at once.
  • The rete alpha registry is synchronized. It is JVM-lifetime shared state reached from the store observer hooks, which fire on whichever thread is writing, and a HashMap racing its own rehash can leave a reader spinning on a probe loop that never terminates. Its check-then-put is one step too, so two callers cannot leave the loser's alpha permanently unmaintained.
  • load-source claims the catalog under one monitor. The busy test, the already-loaded test and the registration were three separate reads, so two requests arriving together each passed all three and spawned a loader.
  • The browser reads untrusted EDN under Throwable, as the daemon does. A deeply nested form overflows the reader's stack with a StackOverflowError, which an Exception catch lets escape — a 500 where an unreadable term is the ordinary answer.
  • The index snapshot's roots-fallback blob is validated like the sections beside it. roots-fallback.nippy carries argument-root postings, which are primary index truth, and a missing or torn blob loaded as [] behind a warning while every argument-root read answered #{} out of a snapshot that opened clean. The meta records the blob's count and byte length, and the load thaws strictly.
  • The mapped index snapshot survives a JVM shutdown, and a failed save leaves the previous image intact. The stamp is taken against the records before a byte moves, durability registrants close in phases, and every section lands in a .tmp until the swap. A failed open likewise gives back the handles it took.
  • An export dump carries every provenance stamp, and export → import → export is byte-stable. The provenance walk covers justification handles as well as sentex ones, and import stores a justification's antecedents as a vector, the shape the engine's own write path stores.
  • The JTMS dedup index carries the identity of the TMS it mirrors. A nested chain over a second KB — legal from an :on-progress callback, with overlapping handle spaces — could answer one KB's dedup question out of the other's supports. Keys coerce fixnum boxing to Long at the boundary, since the map compares with Java equals where the scan compares with =.
  • prove-within prepares its goal, through the same prepare-goal-for-read every other read path takes, so a reifiable NAT or a merge-retired spelling is the same question under the bounded prover that it is under ask.
  • The rete forward matcher fans over predicate-genl sub-predicates at every arity, as the reference res/match-pattern does. Fanning only for a two-element sentence gave the opt-in matcher a different belief set on any rule whose antecedent had another arity.
  • A firing refused at derive time comes back when its exception releases. place-conseq does not place a firing whose exceptWhen exception already holds, and such a firing left no justification and nothing in jtms/blocked — so a settle pass could not see it and the conclusion stayed suppressed after the block lifted. The same knowledge in the other order concluded it. The refusal is recorded as [rule handle, bindings], capped at 4096 entries per rule. docs/exceptions.md.
  • Five order-independence repairs. contradictions names the same side of a clash whatever order the two arrived in; the two settle sweeps sharing one exposure-instance budget walk their moved region in content order; query with {:proof? true :portfolio? true} returns each answer once; negation-nogoods writes with a compare-and-set; and the node engine's inline join plans with the :est-override belonging to its registry leaf.
  • A forward rule fires on a claim argument-position preservation licenses, so sentexes-matching and ask stop disagreeing about the same knowledge. (argPreserving largerThan 1 genl) beside (largerThan dog cat) licenses (largerThan chihuahua maine_coon), which ask reached while the fixpoint fired only on the claims that were written — so the conclusion it never drew had no why, no retraction path and no way to be an antecedent. The join contributes the handles the inherited claim was read from, so retracting any of them withdraws the conclusion. One asymmetry is left: a justification confers the weakest class it rests on, so a :monotonic claim declared preserved by a :default declaration draws a :default conclusion. docs/inherit.md.
  • A head-existential rule carrying an aggregate mints a ground witness. skolem/frontier-vars subtracts a post-join literal's output, so the Skolem NAT no longer takes a variable into its argument list.
  • An open disjoint goal is enumerated from the declarations rather than from the vocabulary. A separation convicts two subtrees, so the answers are the subtypes of what a visible declaration names and the cost is the answer's own size; 0.2.0 asked taxonomy/disjoint? once per type, and once per pair with both arguments open. On 4,000 types carrying one separation that is 15.4 ms to 0.13 ms with an argument bound — flat where it grew linearly — and at 1,000 types the two-variable goal goes from 2.5 s to 4 ms. lein perf's disjoint-enumeration check is the claim.
  • A definitional clash is arbitrated from a context that can see both halves. The checks are scoped to the context they are asked in, so a pair whose halves sit either side of a genlContext edge was answerable from exactly one of the two, and only when that half was the one the settle moved. settle/clash-askers runs the check from the candidate's own context and from the maximal common descendant of it and each context holding a sentex it could pair with; nothing is widened.
  • A pair per opposing sentex, not per opposing type. One sentence stated in a general context and again in one that sees it is two sentexes, of possibly different strength, and a claim that denies it denies both — where the checks named one handle each, so the content-first of the two was weighed and the other left believed beside content that contradicts it.

0.2.0 — 2026-08-03

Not a drop-in upgrade from 0.1.0. Several of the changes below refuse input 0.1.0 accepted or change an observable contract — each such entry is marked Breaking — which is why this is 0.2.0 and not 0.1.1. Entries between here and the 0.1.0 header are in it, newest first.

  • The argument roots are scoped by predicate ([:argument-root pred pos term]), so a materialising join reads one literal's postings rather than wading through every functor's at a shared slot. An [:argument-slot pos term] roster, reference-counted off those postings, keeps the predicate-agnostic reads answerable as a union over the predicates present. The packed long has no room for a fourth key part, so the dense roots route the family to their boxed fallback. index-layout-version is 2: an index written by 0.1.0 reads as :layout-changed and is rebuilt on first open — no action needed, but a large durable store pays a reindex for it.
  • Breaking: every handle-taking fn refuses a non-handle (:bad-handle) — the vector assert returns for a conjunctive rule included, which 0.1.0's retract! silently answered with {:removed-sentexes 0}. nil stays a question with an answer (in? false, why {:stored? false}, add-provenance a no-op), and check-edit reports what edit! throws. why also takes {:max-depth n} (default 256), marks a capped branch {:truncated? true} instead of overflowing, and refuses bad opts (:unknown-option).
  • close! releases a durable KB's directory without waiting for JVM exit, and import! is export!'s inverse. An unclean close still releases the lock and registry; the first component failure is rethrown after.
  • An argIsa / interArgIsa / argGenl refusal names its convicting declaration in content order, not in whichever order retrieval enumerated.
  • The five sweeps run in CI — dense TMS, incremental matcher, node query engine, its tacticians, reference retrieval — each failing-set-identical with the default it replaces. Nothing ran them before.
  • Breaking: assert refuses a non-map opts (:unknown-option) — (assert kb s ctx :monotonic) stored a defeasible sentence in 0.1.0. check already reported the same request; the two agree now.
  • Breaking: open-kb recovers by default (:recover? :auto). The old :warn default handed back a KB that answered wrongly from a reopened store. The cost moves to construction — O(records) on a populated store — and {:recover? false} defers it. :warn and false remain.
  • The public surface is six namespaces, where it was one: vaelii.core plus thin shims vaelii.client, vaelii.starter, vaelii.web, vaelii.serve and vaelii.cli over the impl namespaces they front. The boundary is now what the docs said it was.
  • Breaking: vaelii.client's assert and assert-rule are spelled bare, without the ! 0.1.0 gave them. A ! marks a fn that destroys stored knowledge and neither does — both are additive, and retract! is what takes one back — so the client now spells them exactly as vaelii.core does. A call site writing c/assert! or c/assert-rule! no longer resolves.
  • A clash names the sentex that states the membership, not one that merely entails it, under either retrieval strategy.
  • An auto-compaction queued for a closed store is dropped, so the next open no longer replays it as a crash-interrupted compaction.
  • Breaking: error :type keywords are plain across the tree (:unknown-source, not :vaelii.impl.catalog/unknown-source), and open-kb's backend refusals carry one. Swept in the same pass: the settle re-check queue no longer drops entries queued by a concurrent thread, and foreign/register refuses with ex-info rather than an elidable {:pre}.
  • Leiningen 2.10 is the minimum: :preserve-eval-meta needs it, and 2.9 ignores the key silently.
  • Breaking: POST /op requires Content-Type: application/edn. The type is not CORS-simple, so a browser must preflight and the daemon answers no CORS headers — which closes cross-site request forgery against a loopback daemon. A client that sent no content-type is refused; add the header.
  • Breaking: a request body over 16 MiB is refused (413) before it reaches the heap; VAELII_MAX_BODY_BYTES adjusts the cap.
  • Breaking: DNS rebinding is closed on both servers. Every route requires a Host naming the interface the server was started on. A request with no Host still passes (a non-browser client carries no ambient browser context); a reverse proxy or local alias sets VAELII_ALLOWED_HOSTS.
  • +with-foreign names a coordinate that exists (com.vaelii/vaelii-foreign); the bare id it carried resolved nothing.

0.1.0 — 2026-07-31

The first release. What follows is the development log that produced it, newest first; every entry below is in 0.1.0.

2026-07-30

  • A declaration re-checks the exceptions it moves: (symmetric P), (transitive P), (inverse P Q) and the argPreserving forms change what may be concluded with no fact arriving; (functional P) sweeps the extent when it lands.
  • An equality restates a fact for each reader rather than once for the KB, and is itself a re-check trigger for exceptWhen, unknown and census reads.
  • A change feed: the region a settle already computes is handed to a listener instead of discarded.
  • English in — a sentence read into candidates a person still has to accept.
  • A qualitative relation two contexts entail together fires a forward rule; a believed negative reaches the wiring a positive does; "some context" means the union of what the readers answer.
  • Three readers of one question agree over a cyclic hierarchy, and settle repairs the context ranking after reconciling belief as well as before it.

2026-07-29

  • One front door for backward chaining: the four paths measured, then consolidated to two chainers behind one entry point with one dial. A proof of an ephemeral answer reads the way why does.
  • The goal frontier's order is a policy measured on time-to-first-answer, :ground-first by default; the goal-stack chainer drives one solution at a time and level 7 streams its search.
  • Foreign formats arrive as a classpath plugin, so a reader ships and retires without touching the engine.
  • A constraint declaration may name a second sentex it must not weigh, and a depth bound has no default because there is no defensible one.

2026-07-28

  • lein gate: lint, the suite, and the scaling claims, measured and failed on rather than asserted; five checks added for costs that grow with what they must not.
  • The columnar index is written once and mapped back; backend names read <records>-<index>, all seven.
  • A literal's matches are remembered and retired on a clock; converging branches share one rule expansion; a third backward chainer whose state is a frontier.
  • The naming invariants belong to the knowledge base, and the bulk door counts what it skips.

2026-07-27

  • Aggregation: a count is a query operator, and a firing that rests on one is maintained like any other — gated by a permutation test. A census counts distinctness through the representative the asking context elects.
  • argIsa entails as well as constrains, behind a toggle, retroactively too.
  • A definitional clash names a second sentex, making it a nogood, and the arbitrating sweep asks the taxonomy rather than a fixed functor set.
  • The qualitative network lives on the knowledge base and warm-starts off its own previous answer; a violation ledger is a claim about one KB.
  • The browser draws term shapes, composes English at three densities, and gained lein browser; OpenCyc loading went from 378s to 277s.

2026-07-26

  • Contexts got a vantage: every taxonomy supporter records the context it asserts from, so disjointness, matching fan-out and settle all read only what the asking context can see.
  • A firing names the genl edges it subsumed through — belief and strength run through them like any antecedent, checked against all 24 orderings.
  • Records and index became two independent choices, plus an overlay backend: a private writable fork over a shared read-only base.
  • A knowledge base is readable before it finishes loading, and the suite runs on every backend from one script.

2026-07-25

  • OpenCyc, read and re-expressed: every constant given back its role, 1.1M sentexes in the engine's own format, on the machine that reads it. Nothing of Cycorp's is redistributed.
  • An export format no rename can break, with xz, an importer, and an oracle comparing two knowledge bases; a dump lands every record at its handle.
  • Qualitative spatial and temporal reasoning: RCC-8 and three more spatial algebras, Allen intervals, durations and instants, behind one glue.
  • One gap written in two units is one constraint: both spellings snap to the tolerance grid, and a unit given two conversion factors is its own base.
  • A knowledge-base catalog with a browser that switches between KBs; inherit declared rather than assumed; definitional checks reach every term; argGenl constrains one level up.

2026-07-24

  • The scale program opened with measurement first: the truth-maintenance wall, a posting-encoding bake-off on a real corpus, and a rule audit.
  • Three dense representations, each measured: :memory-dense integer postings, the :memory-columnar int-token trie with CSR compaction (3.18x whole-index), and a bitmapped TMS behind a protocol.
  • The disk side got dense too: positional record reads with a hot-record LRU, a positional frame codec, tokenized bodies over a durable dictionary — all behind a backend parity gate.
  • Recursive forward chaining went O(n³) → O(n log n); a term roster enumerates vocabulary in O(terms).
  • The web browser was hardened — escape by default, guard the parse, bind loopback, refuse cross-origin writes — and a pluggable LLM proposes edits and never applies them.

2026-07-23

  • A performance review, its findings fixed: the disk log records operations rather than grown values (killing an O(N²) write amplification), settle keeps a coincidence set, re-checks narrow to the moved cone, region fixpoints became semi-naive worklists, compaction is copy-on-write.
  • Symbolic equational reasoning: pure oriented rewriting with full Knuth-Bendix orientation, order-independent normalization, and non-confluence surfaced; rewriteOf extended over predicates and types.
  • An operational surface: an EDN-over-HTTP daemon, a command-line driver, a thin client, and a browser that attaches to a daemon — with the public API closing every hole the browser tracked.
  • Existential rule heads with deterministic skolemization; an occurs check; closures answered on demand; the record split into atomic and rule shapes with interned symbols; a bulk-load fast path.

2026-07-22

  • Negation as failure, at top level and in antecedents, with block, sweep and revive, and stratification to keep it sound.
  • Resource-bounded anytime inference with qualitative cost tiers and a cost-ordered forward-chain join; ask-within normalizes its goal.
  • Reification of non-atomic terms; structural subterm indexing, oracle-proven, then on by default.
  • The storage seam: index logic onto one key-value protocol, an in-memory backend, then the on-disk substrate — files, lock, durability, record store, index store.
  • The index benchmark harness, and a per-handle provenance side map.

2026-07-21

  • exceptWhen canonicalized into the record, blocking excepted conclusions with only reachable firings re-checked; its query reified the way a fact is.
  • Equality landed: the closure, the different prover, a specification suite, and wiring into assert; stratification is checked on edge change.
  • The engine split out of one namespace into five, and the knowledge base restructured into a layered tree loaded on start.
  • assumptionRules with persistent solve and labeling contexts, proven on a sudoku.
  • Retrieval got sharper: argument roots, multi-column narrowing, predicate subsumption, set-algebra retrieval, an opt-in incremental matcher.
  • Truth-maintenance mutations are atomic; lint arrived.

2026-07-20

  • Canonical rule form — canonical variables with a varmap, literal order, comparison direction — so rules alike up to renaming share one handle.
  • The eight-level lookup-to-query stack, lazy throughout, with a browser page showing which level answered.
  • Order independence and locality pinned as invariants: region-local relabelling, belief-following closures, content-keyed tie-breaks.
  • The answer-set layer wired to the edge-solver seam, with a labeling materialized as a context; the defeasible layer made sound, six bugs pinned as failing tests first; exceptWhen began as a failing suite.
  • Everything but core moved under vaelii.impl.*; ! reserved for irreversible operations; tests became net-neutral, and a second concurrent run fails fast rather than corrupting the first.

2026-07-19

The first day: a contextualized common-sense knowledge base with a trie index, inference and truth maintenance.

  • Sentexes — a sentence plus the context it holds in — stored as records behind protocols with nippy serialization; rules are sentexes too, with built-in transitivity for types and contexts.
  • Forward chaining with dependency-directed retraction, and a backward chainer; a non-monotonic TMS with strengths, soft prioritized contradictions, and a solver seam.
  • An inverted term index, directed rules, disjointness, well-formedness checks, and a pluggable prover query engine; structural connectives canonicalize into the record; evaluable arithmetic.
  • A web browser over the whole thing, over a starter ontology with every term documented.

Can you improve this documentation?Edit on GitHub

cljdoc builds & hosts documentation for Clojure/Script libraries

Keyboard shortcuts
Ctrl+kJump to recent docs
Move to previous article
Move to next article
Ctrl+/Jump to the search field
× close