circagraph
Sign inSign up

openchronology/timeline-app on GitHub

The OpenChronology repository on GitHub, mirrored by och-import-github: issues and pull requests as durations (ongoing while open), what happened to them as moments, plus releases, milestones, commits and branches. The Branches view shows the most recently active branches each in a track of its own.

Source: https://github.com/openchronology/timeline-app

3,131 moments · 2026-10-05 CE 20:21:49 UTC – 2026-11-16 CE 12:10:45 UTC · by @athan · @athan

#392 Docs disagree with the code in a dozen places (found while writing disclosures)

–

  • github
  • github/openchronology/timeline-app
  • github/openchronology/timeline-app#392
  • issue
  • state:closed

While writing the technical disclosures (#388) these places were found where the docs disagree with the code. The disclosures follow the code; the docs below still need fixing. - `docs/architecture.md:120` says saving is still a full rebuild and incremental index writes are future work. That contradicts the "Incremental saves" section of the same file (L46–59) and `server/incremental.mjs`. - `docs/architecture.md:57`: a patch changing more than half a timeline is rebuilt — but any patch of 64 changes or fewer is applied in place (`incremental.mjs:78-87`), so a rebuild needs size > max(64, count/2). - `docs/durations.md:46`, the OpenAPI text and a comment in `store.mjs` say PostgreSQL takes the first 10,001 durations in start order. The query (`store.mjs:2118-2126`) has `LIMIT 10001` with no `ORDER BY`, so it takes them in walk order (SQLite does keep start order). Decide whether to fix the query or the docs. - `docs/architecture.md:38`: `rasterize` applies "only after gestures". `navigate()` (`src/app.ts:4641`) also rounds when fitting a duration, jumping to a search result, opening a shared view link, on every zoom-animation frame and for follow-latest. - `docs/architecture.md:63` names `oc_overview`; the server calls `oc_overview_v2` (`oc_overview` is a wrapper). - `docs/architecture.md:69`: rebuild-saves compare the whole timeline "in one query"; the code pages through 20,000 keys at a time (`history.mjs:443-547`). - `docs/architecture.md:74`: a copy is prepared later at "20,000 events"; the code counts the source's own moment rows plus duration-tree rows, hidden rows included (`store.mjs:549-557`). - `docs/architecture.md:80`: history starts at the first save after the upgrade; the migration also starts it (`migrate.mjs:42`). - `docs/architecture.md:106`: resolution levels are exact powers of two; above ~8,192 px of drawable width the level is 1.5·span/1024 (`remote-cache.ts:160-166`). The last-drawn window also stays on screen whenever it overlaps the view, even without covering it (`remote-cache.ts:249-252`). - `docs/time-presentation.md:71`: two ruler labels crossfade; since #241 only the heaviest label is drawn (`ruler.ts:453-477`). - `docs/plugins.md:260`: "64 MiB" and "compiled once per process"; the code sets 64/16/16/4 MiB limits and remembers at most 1,000 results (`plugin-sandbox.mjs:20-27, 129-142`). - `docs/log-import.md` (~L110): batches grow to 5,000 entries; the importer caps them at 4,999 (`import-logs.mjs:258`). Code-only leftovers: - `durationLanes` is unused for drawing. - The `'size'` marker kind is in the type but rejected by validation. - `src/custom-time.ts:12` says no source text reaches the server runtime; the server compiles it but never runs it. Line numbers are as of `007091c`. https://github.com/openchronology/timeline-app/issues/392

Related

Open #392 Docs disagree with the code in a dozen places (found while writing disclosures) in the timeline

#392 Docs disagree with the code in a dozen places (found…