DO-1 (regression): the orderedList start hardening `Number(x)||1` let a
negative/fractional start through into the marker ("-3.", "2.5.") — marked
can't tokenize it, so the whole list re-imported as a paragraph (structure
corruption); start=0 churned. Both the markdown and raw-HTML export paths
now use `Number.isInteger(raw) && raw > 1 ? raw : 1`, so any degenerate
start collapses to the default "1." markers / bare <ol> (list always valid).
New ordered-list-start-normalization test pins {0,-3,2.5} → valid start=1
list, byte-stable; the round-trip fuzz keeps to integers >=2 (num 2,3,5,42).
DO-2 (nightly was non-functional): NUM_RUNS=5000 OOM'd the worker and the
crash was misreported as a "counterexample" issue whose prefix-dedup then
locked out all future issues. Reworked to shard 8 fresh vitest processes
(600 runs each, distinct seeds, --max-old-space-size) so deep fuzzing
never OOMs; a failing shard's output is preserved, and issue creation
discriminates a real fast-check counterexample from an infra/OOM failure
(distinct titles + scoped dedup). The two issue steps use `always() &&`
so they actually run on the failure path.
DO-3: envInt extracted to test/generative/env-int.ts + unit-tested.
DO-4: nightly dispatch inputs go through env: (no ${{ }} in run:).
DO-5: attr-arbitraries.ts docblock synced (column.width/orderedList.start
are fixed+fuzzed, not pinned it.fails).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Items 1 & 2 of the #351 remainder.
Item 1 — the flat/nested generative property tests now read SEED and
NUM_RUNS from PROPERTY_SEED / PROPERTY_NUM_RUNS (via an envInt helper that
honors an explicit 0 and falls back to the current defaults —
20250705/300 flat, /100 nested — on unset/empty/non-numeric). New
.github/workflows/nightly-property.yml runs the generative suite daily
(and on workflow_dispatch) with a random seed and NUM_RUNS≈5000; on
failure it files a Gitea issue containing fast-check's shrunk
counterexample (jq-escaped, dedup'd by title). No build step — the suite
imports the converter from src/, so a tsc error can't masquerade as a
property failure.
Item 2 — new packages/prosemirror-markdown/README.md documents the
counterexample process (surface -> shrink -> permanent fixture in
test/fixtures/counterexamples/ + counterexamples.test.ts -> fix the
converter, never weaken a property; maintainer-approved ACCEPTED/allowlist
entries carry a reason) plus the two golden layers, the coverage
allowlist, and how to run with the env knobs. Linked from AGENTS.md.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Item 4 of the #351 remainder — implement value-fuzz for the media
dimension/family attributes previously parked in the flat property
suite's ATTR_VALUE_FUZZ_ALLOWLIST as "round-trip candidates deferred to
a later PR": width/height/align/aspectRatio/size/caption/title/alt across
image, video, youtube, pdf, drawio, excalidraw, embed.
Each gets a non-default-value arbitrary in attr-arbitraries.ts and is
removed from the allowlist; the P1 (semantic) + P2 (byte-stability)
property gate now exercises them. All round-trip green — the comment-JSON
serialization (stable key order, String()-stringified) carries every one
faithfully, so no new pins or accepted-limitations were needed.
embed.width/height are fuzzed as numeric STRINGS (not numbers): a
non-default embed dimension round-trips through the comment JSON as a
string (the numeric 800/600 default is still omitted/re-materialized),
so authoring a number diverged under P1.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Item 3 of the #351 remainder — resolve the two pinned converter
counterexamples, fixtures-first.
orderedList.start (genuine P1 loss): the converter always emitted "1."
and dropped attrs.start. Now the markdown path emits `${start + index}.`
(Number-coerced, guards a stray non-numeric start) and the raw-HTML path
emits `<ol start="N">` when start>1; a default (start=1) list is
byte-unchanged. tiptap StarterKit reads both forms back. Un-pinned as a
passing regression test; orderedList.start is now value-fuzzed.
column.width: the "50% churn" counterexample rested on a false premise —
the canonical editor (editor-ext column.ts) stores width as a unitless
flex-grow NUMBER (parseFloat, `flex:${width}`), never a "%" string, and
docmost-schema.ts is a vendored mirror that MUST match it. The original
parseFloat was correct parity and already byte-stable for numeric widths.
Reverted the mirror to parseFloat, deleted the fabricated counterexample
fixture, and now value-fuzz column.width as a number (real type). No src
behaviour change for column.width — parity with editor-ext preserved.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Финальная ступень #351: генератор ЦЕЛЫХ вложенных документов (random walk
по ContentMatch схемы, min-depth fixpoint для терминирования, глубина/размер
ограничены) + инвариант P4 (фазз парсера: для ЛЮБОЙ строки markdownToProseMirror
не бросает и результат валиден по схеме). Инварианты P1 (семантический
round-trip), P2 (байтовый fixpoint со 2-го прохода), P3 (тотальность) — строгие.
Генератор сразу нашёл классы реальных багов конвертера; все починены (инварианты
НЕ ослаблялись — чинился конвертер):
- loose (много-блочные) контейнеры (listItem/taskItem/callout/detailsContent)
склеивали блоки при реимпорте — ТИХАЯ ПОТЕРЯ ДАННЫХ; теперь blank-line
разделитель между блок-детьми (по образцу blockquote).
- соседние sibling-списки одного marker-family (task+bullet и т.п.) сливались
в один список с ПОТЕРЕЙ чекбокса — теперь между ними эмитится инертный
`<!-- -->` разделитель (byte-stable round-trip).
- paragraph textAlign терялся во вложенных li/td/th.
- вложенный codeBlock терял хвостовой перевод строки.
- pageBreak/pageEmbed/subpages/transclusion дропались во вложении
(blockquote/callout/details/li).
- медиа в columns: number→string ширины/высоты и лишний data-align (churn).
- callout `> [!type]`, вложенный в список/цитату, парсился неверно
(prefix-aware regex).
Golden-обновления (6) — прямые следствия loose-container фикса, каждое
round-trip'ится. 2100+ сгенерированных документов (3 seed) — 0 падений P1/P2.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
F1 [WARNING] The 'no invisible coverage hole' guard enumerated only
schema.nodes, so MARK attributes silently escaped the value-fuzz completeness
check — link.internal/target/rel/class are never fuzzed and nothing flagged it,
and a new attributed mark would slip through. Added allSchemaMarkAttrKeys() plus a
MARK_ATTR_FUZZED / MARK_ATTR_ALLOWLIST registry and two tests: every schema mark
attr must be in exactly one set (a new one turns it red), and neither set may hold
a stale row.
F2 [WARNING] The ACCEPTED annotation misclassified table colspan/rowspan as
having 'no md representation'. They DO round-trip — a spanned cell makes the
converter emit the whole table as a raw <table> with colspan/rowspan, which the
tiptap parser reads back. They are frozen only because generating a
geometrically-valid spanned table is deferred PR-2 structural work (the flat
generator hardcodes span = 1), not a markdown limit. Reclassified them as
DEFERRED-BUG (distinct from ACCEPTED) so a maintainer does not read them as an
inherent limitation; colwidth / backgroundColor(Name) stay ACCEPTED (the
raw-<table> fallback drops them).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Schema-derived, property-based (fast-check) round-trip tests over flat
single-node ProseMirror documents. One test PR — src/ is untouched; the two
real bugs found are pinned as loud it.fails counterexamples, not fixed here.
- attr-arbitraries.ts: per-attribute four-state arbitraries (absent/default/
nonDefault/degenerate), attribute list sourced from schema.nodes[t].spec.attrs;
a documented override table supplies legal domains for constrained attrs and
distinguishes two frozen classes explicitly — ACCEPTED limitations (no md
representation) vs PINNED bugs (representable but dropped, tracked as
counterexamples).
- text-arbitraries.ts: hostile text corpus (ported from the existing property
test's supported-space guarantees).
- node-generators.ts: flat single-node generators + a completeness contract —
every one of the schema's 45 nodes / 12 marks is either generated or listed in
KNOWN_UNCOVERED with a reason.
- flat-roundtrip.property.test.ts: P1 (semantic round-trip via
docsCanonicallyEqual), P2 (second-pass byte fixpoint — anti GS-EDIT-REVERT),
P3 (totality), generator validity via schema.check(), and an explicit
attribute-value-coverage snapshot so the not-fuzzed set can never grow silently.
- counterexamples: column.width (% dropped on parseFloat -> P2 churn) and
orderedList.start (non-1 start renders as '1.' -> P1 loss) pinned as it.fails.
SEED=20250705, NUM_RUNS=300 per property; ~17s, no OOM (union arbitraries).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
agent_coder
agent_coder