Files

T

m038andClaude Opus 5 7c9c140a1b Merge main into feat/docs-reconcile; defer to main on two overlapping fixes

main advanced 13 commits while this documentation audit ran — the
location-override work was merged into the outer repo — and it independently
fixed two of the audit's findings. Without this merge the branch would have
REVERTED them, which is the worst outcome for a cleanup pass because it arrives
disguised as an improvement.

Both conflicts resolved in main's favour; main's wording was better informed:

- CLAUDE.md, the single-map-path rule: main's carve-out (829325c) states the
  exception as its own top-level bullet, names MAP_STYLE as the one shared
  thing, and spells out both prohibitions ("do not fold it into initEntryMap",
  "do not add a third path"). Taken verbatim over the version drafted here.
- 2026-07-23-post-form-location-override.md: main (a517331) had already set the
  status to Complete, with far richer detail — the multi-agent review findings,
  the green-run numbers, the DEL4 regression still open, and the merge SHAs.
  Taken in full; the audit's claim that the status "lagged" was dropped, since
  it was true only of this branch's older branch point.

Submodule pin: main bumped user/ to dd19995 and this merge preserves that. The
audit's own no-gitlink-commit discipline applies to bumping the pin as a side
effect of routine work, not to discarding a bump main already made.

main touched none of the other nine corrected documents, so the remaining 18
findings stand unchanged.

Audit notes corrected to match reality rather than left overstated:

- superseded-decisions.md R13 now dates the carve-out to 2026-07-24 (829325c)
  rather than implying this pass introduced it.
- The reconciliation spec gains an "audit baseline moved twice" section: the
  submodule pin lagged real HEAD, and then the base branch advanced mid-audit.
- The compounded learning's section 6 is rewritten from "audit the current
  state" to "re-check the baseline before publishing, not only before starting",
  with the two habits that actually follow: merge the base branch in before
  opening the PR and read conflicts as findings, and when the incoming version
  is better, take it wholesale. An audit has no special authority over the work
  it audits.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

2026-07-25 00:21:47 +02:00

6.7 KiB

Raw Blame History

Docs Reconciliation — Design

Date: 2026-07-25 Status: Implemented

Reconcile the documentation against the code after five weeks of undocumented evolution, so that a repeat review returns "ok".

Problem

Documentation for this project began as thoughts and plans. The app then changed — features were built differently, some were dropped, and the owner changed his mind about what he needed. Those decisions were recorded ad hoc or not at all. The result is a tree where some docs describe a site that no longer exists, and nothing marks them as historical.

Concretely, before this pass:

docs/working/README.md advertised summary.md as the project's current state, while summary.md described Leaflet, a /tracker feed, a /map page, a /stats page, and a "Journal · Map · Stats" nav — none of which exist.
docs/reference/design-system-light.md documented a light-mode palette in present tense. No light mode is implemented anywhere: tokens.css has a single :root block and no prefers-color-scheme / data-theme mechanism.
CLAUDE.md — the always-loaded file — asserted a source relationship that does not exist (css-compiled/ generated from css/style.css + css/tokens.css).
README.md's server runbook documented every make remote-* command without the -test/-prod suffix that guard-env requires, so the documented commands cannot run.
docker-compose.yml still defines a travel-memories service whose source was deleted in a80b0a9 ("moved to separate project"), so make start fails on any clean checkout.

Root cause

Per docs/solutions/conventions/claude-md-content-tiering.md, descriptions drift because the code moves and the prose does not; rules do not drift, because they encode intent rather than state. This pass confirms that finding again: every defect found was a description of code, config, or a command — not one was a rule that had become wrong on its own.

The compounding factor is tense. The tree mixes two kinds of document with no marker distinguishing them:

Kind	Files	Staleness is
Present-tense — "this is how it is"	`CLAUDE.md`, `reference/`, `guides/`, `README.md`, `CONCEPTS.md`	a defect
Past-tense — "this is what we decided then"	`working/plans/`, `working/specs/`, `working/milestones/`, `summary.md`, `pm-analysis.md`	correct and expected

A completed plan should be stale — it is a record. It only becomes a problem when nothing tells a reader it is a record. milestones/milestone-2.md opens by describing a Leaflet /map page in confident present tense with no date qualifier.

Approach

Two mechanisms, combined:

A — one authoritative supersession ledger. docs/reference/superseded-decisions.md records every reversal in one table: what was planned, where it was planned, what is true now, when it changed, and why. This answers "what did I change my mind about?" in a single place, which is the question a review actually asks.

B — inline notes at the point of staleness. Every superseded section carries a > **Superseded …** blockquote where the stale claim sits, so the claim can never be read un-corrected. This pattern is not invented here — docs/reference/architecture.md and docs/guides/trip-switching.md already use > History: and > **Changed 2026-07:** notes.

A alone has an indirection problem (a pointer you may not follow). B alone has a completeness problem (no changelog view, and coverage is only as good as the annotation pass). Together each covers the other's gap.

Scope, split by tense

Present-tense docs are corrected against the code. The code is the source of truth. Every factual claim was verified by reading the code, config, or Makefile — not inferred.
Past-tense docs are annotated only, never rewritten. 41 plans and 25 specs, ~30k lines. Their ✅ Complete trailing notes are good records; rewriting them would destroy the audit trail and is unbounded work.
Code-side inconsistencies are logged, not fixed. Mixing behaviour changes into a documentation diff would make it unreviewable. They go to docs/working/2026-07-25-doc-drift-recommendations.md for a separate decision.

Out of scope

A repeatable drift check (script with an exit code). Deliberately deferred — the owner asked for the one-time reconciliation first. It is the lead recommendation in the recommendations doc.
Fixing the travel-memories / docker-compose.yml breakage, the unused shortcode-gallery-plusplus, and the italy-2025 demo fixtures. All logged as recommendations.

Verification

Claims were checked against, not assumed from:

Claim area	Verified against
Nav labels	`templates/partials/base.html.twig:27-31`
Template + partial inventory	`ls templates/`, `ls templates/partials/`
Asset sources → outputs	`user/themes/intotheeast/package.json` build script
`css-compiled/` provenance	CSS imports in `js/src/*.js`; `assets.addCss` in `base.html.twig:7-8`
Design tokens	`css/tokens.css`
Light mode	absence of `prefers-color-scheme` / `data-theme` and of light hex values in `css/`
Photo field rules	`user/pages/02.post/post-form.md:35-46`
`hero_image` removal	`post-form.md:149-151`
`entry-actions` routes	`user/plugins/entry-actions/entry-actions.php:63-73`
`make` targets + env guard	`Makefile` (`guard-env:41-43`, `make-env-target:45-46`)
`travel-memories` removal	`git log -- services/` → `a80b0a9`; `docker compose build` failure

The audit baseline moved twice

Both times, auditing the convenient state rather than the real one would have produced wrong findings.

The submodule pin lagged. A fresh worktree checks out the user/ commit the outer repo pins, not user/'s real HEAD. The pin predated the merged location-override work, so auditing it would have reported a feature as unbuilt and missed two new source files. user/ was moved to its real HEAD (dd19995) before auditing, and the gitlink deliberately not committed.

The outer main advanced 13 commits mid-audit. The location-override branch was merged into the outer repo while this pass was running, which independently fixed two of the findings — the single-map-path carve-out (829325c) and the plan's Status: line (a517331). Merging main in before opening the PR was what surfaced that; without it this branch would have reverted both. main's wording was better than the replacement drafted here and was kept in full. main touched none of the other nine corrected documents, so the remaining findings stand unchanged.

The general rule: re-check the baseline before publishing, not only before starting. A long audit races the work it is auditing.

6.7 KiB Raw Blame History