refactor: sitemap throttle + SEF URLs (#100) and batch cursor pagination (#106) #116

Merged

jmiller merged 1 commits from feat/sitemap-batch-redesign into dev 2026-06-29 17:18:12 +00:00

Conversation 0 Commits 1 Files Changed 4

+85 -16

jmiller commented 2026-06-29 17:17:59 +00:00

Owner

Copy Link

Copy Source

Implements the two deferred redesigns. 4 files, all php -l clean.

#106 — Batch cursor pagination

BatchController::process() previously always queried offset 0 and relied on IS NULL to skip done rows — so a row that failed to insert was re-fetched on every chunk forever, and the created > 0 stop-guard masked it by halting the batch early.

Now it paginates by id cursor (WHERE c.id > :lastId ORDER BY c.id):

• Failed rows fall behind the cursor and are never revisited → the loop is provably terminating
• process() returns examined + last_id; the editor JS passes the cursor each chunk and stops when a chunk examines 0 rows (reaches 100% even with failures)

#100 — Sitemap throttle + SEF URLs

(The security part — public-access filtering + atomic write — shipped earlier this cycle. These are the remaining two items.)

• Throttle — onContentAfterSaveRebuildSitemap regenerates at most once per 60s (SITEMAP_MIN_INTERVAL), so bulk edits/imports don't rebuild the whole sitemap on every save. Eventually consistent within the window.
• SEF URLs — each article is routed via Route::link('site', …) so the sitemap matches the canonical URLs the plugin emits, with a try/catch fallback to the non-SEF index.php URL if routing fails. Worst case = today's behavior; never a broken URL.

⚠️ Verification note

Static-only (php -l). Two things warrant a runtime check on Joomla 6: that Route::link('site', …) produces correct SEF URLs from the admin/save context (else the fallback kicks in), and the cursor batch reaching 100% on a site with eligible articles. Both are designed to fail safe.

Remaining minor follow-up: per-language sitemaps (optional).

Closes #100, #106.

🤖 Generated with Claude Code

Implements the two deferred redesigns. 4 files, all `php -l` clean. ## #106 — Batch cursor pagination `BatchController::process()` previously always queried **offset 0** and relied on `IS NULL` to skip done rows — so a row that failed to insert was re-fetched on every chunk forever, and the `created > 0` stop-guard masked it by halting the batch early. Now it paginates by **id cursor** (`WHERE c.id > :lastId ORDER BY c.id`): - Failed rows fall behind the cursor and are never revisited → the loop is provably terminating - `process()` returns `examined` + `last_id`; the editor JS passes the cursor each chunk and stops when a chunk examines 0 rows (reaches 100% even with failures) ## #100 — Sitemap throttle + SEF URLs (The security part — public-access filtering + atomic write — shipped earlier this cycle. These are the remaining two items.) - **Throttle** — `onContentAfterSaveRebuildSitemap` regenerates at most once per **60s** (`SITEMAP_MIN_INTERVAL`), so bulk edits/imports don't rebuild the whole sitemap on every save. Eventually consistent within the window. - **SEF URLs** — each article is routed via `Route::link('site', …)` so the sitemap matches the canonical URLs the plugin emits, with a **try/catch fallback** to the non-SEF `index.php` URL if routing fails. Worst case = today's behavior; never a broken URL. ## ⚠️ Verification note Static-only (`php -l`). Two things warrant a runtime check on Joomla 6: that `Route::link('site', …)` produces correct SEF URLs from the admin/save context (else the fallback kicks in), and the cursor batch reaching 100% on a site with eligible articles. Both are designed to fail safe. Remaining minor follow-up: per-language sitemaps (optional). Closes #100, #106. 🤖 Generated with [Claude Code](https://claude.com/claude-code)

jmiller added 1 commit 2026-06-29 17:17:59 +00:00

refactor: sitemap throttle + SEF URLs (#100) and batch cursor pagination (#106) ... 7532446e46

#106 — BatchController now paginates by id cursor (WHERE c.id > :lastId)
instead of always querying offset 0. A row that fails to insert falls behind
the cursor and is not re-fetched, so the batch always terminates and reaches
100% even with persistent failures. process() returns examined + last_id; the
editor JS drives the cursor and stops when a chunk examines 0 rows.

#100 — Sitemap:
- Throttle: regenerate at most once per 60s on content save (SITEMAP_MIN_INTERVAL)
  so bulk edits/imports don't rebuild the whole file every save
- SEF URLs: route each article via Route::link('site', ...) with a fallback to
  the non-SEF index.php URL if routing fails (worst case = prior behavior)
(access-level filtering + atomic write were done earlier in the cycle)

jmiller merged commit cb0e5596ea into dev 2026-06-29 17:18:12 +00:00

jmiller deleted branch feat/sitemap-batch-redesign 2026-06-29 17:18:12 +00:00

jmiller referenced this pull request 2026-06-29 17:18:27 +00:00

Minor correctness bugs: loadArticle null-cache, zero-date OG timestamps, batch JS premature halt #106

jmiller referenced this pull request 2026-06-29 17:18:46 +00:00

Sitemap regeneration is unbounded, runs on every save, and leaks non-public URLs #100

Sign in to join this conversation.

Reviewers

No Reviewers

Labels

Clear labels

bug

Something is broken

documentation

Documentation improvements

enhancement

production-readiness

Required before v1.0 release

security

Security vulnerability or hardening

testing

Test coverage and QA

No labels

Priority -

Type -

Milestone

No items

No Milestone

Assignees

Clear assignees

jmiller (Jonathan Miller)

No Assignees

1 Participants

Notifications

Subscribe

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: MokoConsulting/MokoSuiteOpenGraph#116