Reviving Praderas (Day 10) — what shipped today and why
This is a daily log entry: not a long tutorial, but transparency on a concrete block of work (code + content + product judgment) and the clock numbers we committed to on Day 9.
Published changes (executive summary)
- Translation batch 2 — All 13 chapters of Control de Tiempo Desacoplado now have ES/EN pairs (
Translation_Keypraderas-ctd-01…13). The English series display name is Decoupled time tracking, with the sameSeries_Slugso series URLs stay stable. - Missing English hubs — Added
/en/seriesand/en/categoriaspaired with the Spanish pages (navigationTranslation_Keys). The EN navbar no longer dumped readers onto Spanish-only hubs: that was a real product hole, not polish. - Series routes per language — The series plugin understands
/en/series/<slug>/, builds the collection grid for the active page language, and makes the sidebar “full series index” link use the correct prefix (/en/series/...vs/series/...). - Honest EN category counts — On
/en/categorias, counts includeblog/en/posts only. Tag labels remain the shared taxonomy (explicit until we add EN display names). - Templates —
series.twigandcategories.twigbranch copy and breadcrumbs by language;nav.twigtargetsen/seriesanden/categoriason the EN branch. - Small fix — Repaired a broken
</divinside a code block in the Spanish React project-management article (repo hygiene).
Reasoning (why one work block)
- Whole series in one batch — Same rule as Day 9: avoid half-translated chapter paths for readers who follow order.
- Hubs + routing in the same PR — Without EN series routes, the translated arc was still broken in navigation: you could read English posts and then hit Series / Categories in Spanish. Shipping “text only” and leaving that UX gap would have been worse than finishing the cut.
- Language-scoped category counts — Showing whole-archive counts on a page labeled English would be misleading; the neighbors/tag-count path now respects the page language.
Time comparison (~35 min human wall clock)
For this scope (13 post pairs + two hub pages + plugins + templates + tracker + daily log + PR flow), human wall-clock time was on the order of ~35 minutes (review, light manual checks, PR open/merge depending on your workflow).
As an order-of-magnitude comparison with a localization specialist plus classic editorial QA for the same surface area (prose + metadata + internal link sanity + bilingual route checks), a plausible band is still about ~10 to ~18 hours, depending on review depth and whether a TM already exists.
Again: the point is not “AI beats humans,” but what compresses calendar time when the repo already encodes conventions (Translation_Key, batches, glossary) and the PR scope stays tight.
Honest limitations
/tagsand the sidebar search flow remain mostly Spanish-first; the EN categories page notes that. A proper pass is another PR.hreflangon series detail may still need finesse depending on how Pico resolves one template across two URL shapes; the big win was language-coherent navigation and listings.
What comes next
Batch 3 — security, privacy, and geolocation cluster — same batching and glossary discipline.