Implements the 3 HIGH fixes agent_qa found via the RALPH test cycle (analysis theirs,
implementation here), fixtures-first.
- T2E-1 [converter]: a multi-paragraph callout collapsed to one paragraph on round-trip.
markdown-converter `case "callout"` now joins block children with "\n>\n" (a blank ">"
separator), byte-identical to the blockquote serializer, so two paragraphs survive
re-import. Fixture 12-callout-multiblock; two existing tests updated (their old
expectations encoded the collapse bug).
- T6-listnest [converter]: a callout/blockquote nested in a list item corrupted on
round-trip (callout type lost, `[!type]` leaked into text — confirmed DB corruption).
Add bridgeNestedCallouts: a post-marked, nesting-agnostic JSDOM pass that reconstructs
callouts from `[!type]`-opening blockquotes at ANY depth (the indented ` > [!type]`
form the column-0-anchored preprocessor misses), emitting the same callout div as the
top-level path (disjoint inputs, no double-processing). Strict `^[!type]` guard, so a
real blockquote — or one with `[!` mid-line — is NOT converted. Fixture 13-callout-in-list
+ a no-lead control + false-positive controls (proven non-vacuous).
- D-P3-1 [git-sync]: a "ghost" file (a tracked id that was never a page) was silently
absence-deleted on pull-reconcile -> data loss. Add GitSyncClient.pageIdsExist (datasource:
workspace-scoped SELECT over any space incl. trashed/moved, non-UUID ids filtered first);
cycle.ts computes the candidate-delete set and passes existing ids as `deletableIds` into
planReconciliation, which now absence-deletes ONLY real page rows — a ghost is preserved.
A cycle-level e2e test proves the ghost survives runCycle end-to-end.
Suites green: prosemirror-markdown 687, git-sync 272 (no type errors), datasource 36,
orchestrator 26; both packages tsc clean. Scope: the 3 fixes + tests/fixtures only, no
develop re-merge. T4E-esc (LOW escaping) left for the maintainer, per agent_qa.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Closes the architecture item from the #119 review: drop the "vendored from
docmost-sync" framing and the CJS↔ESM `Function('import()')` bridge so the engine
is a normal first-class gitmost package.
Part 1 — vendoring markers removed (prose only, zero behavior change): reworded
"VENDORED into gitmost" / "vendored from docmost-sync" / "Engine LOGIC is
byte-identical" / "it's a port" comments across the engine. Behavior-bearing
strings are untouched: BOT_AUTHOR_NAME/EMAIL and the `Docmost-Sync-Source:`
provenance trailers (changing them would break git authorship + the loop-guard).
Part 2 — the package is now ESM (matching the sibling @docmost/mcp): `type: module`,
tsconfig Node16, `.js` extensions on relative imports, and a static
`import { marked }` replacing the `new Function('return import(...)')` /
`loadMarked` hack — the bridge is GONE from the package. The CommonJS NestJS
server loads the now-ESM engine via a new `git-sync.loader.ts` that mirrors the
existing `docmost-client.loader.ts` mcp loader exactly (Function-indirected
dynamic import + cached promise + retry-on-reject). The 4 server consumers
(orchestrator/datasource/vault-registry/git-http-backend) call `await loadGitSync()`
for value exports; types stay `import type` (erased). The converter-gate spec —
which needs the real converter — loads the package's TS source via a jest
moduleNameMapper + isolatedModules (documented in that spec); the other git-sync
specs mock the loader.
Verified: engine builds pure ESM (no Function/require leftover), vitest 614,
editor-ext build, server + client tsc, full server jest 1397/0. Live stand
smoke-test: server starts clean on the ESM engine (no ERR_REQUIRE_ESM), a real
sync cycle runs through the loader, and the basic e2e suite is 12/12 (clone via
git-http-backend, push, pull, delete, 3-way merge — all through the new loader).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
agent_coder
claude code agent 227