Commit Graph
3 Commits
Author SHA1 Message Date
webmaster 9d14652a57 hide [hidden] site-wide: one [hidden] { display:none !important } in style.css
A filtered-out directory card stayed visible. The card is targeted by an
id-based ready-state selector (`.js #directory[data-dir-state="ready"]
#dirGrid > .dev-card`, two ids), which outranked the old per-element patch
`.dev-card[hidden] { display:none }` — so `card.hidden = true` set the
attribute and every card stayed on screen while the count line correctly
read "Showing 1 of 4".

Fix: one rule, `[hidden] { display:none !important }`, early in /style.css.
A class- or id-based `display` can no longer beat the attribute, so this
cannot recur with the next class that sets `display`. The per-element
`[hidden]` patches are removed so a single mechanism governs:
 - odoo-community-developers: cards + filter row + pager + modal
   (load-bearing — each sets `display`); focus-sliders/noscript (redundant)
 - index/de/es `.hero-notice[hidden]`, odoo-black-friday `.ads-consent[hidden]`
   (redundant — those elements set no `display`)
Re-fingerprint local assets (style.css changed): /style.css?v 3332fbba93 ->
061517baef across 51 pages.

Measured in a real browser (local harness, stubbed webhook; no record created,
no probe write) — computed display + geometry, never the `hidden` property:
 - filtered to one card: the other three report display:none and height 0;
   the visible set (1) equals the count line ("Showing 1 of 4") — 1 == 1
 - HEAD (pre-fix) reproduces the defect: 3 cards hidden=true yet display:flex,
   height 599.38 each; 4 visible vs the count's 1
 - pager: display:none height 0 with one page; display:flex height 66.83 with
   12 cards (Page 1 of 2); page 2 hides page-1 cards (display:none)
 - modals closed: display:none height 0; load state: cards 0, pager/empty
   display:none, filter row visibility:hidden by design
 - 375 px: same result, no horizontal overflow; avatar ask-row top identical
   across a row (2053.97 / links 2086.91)
2026-11-03 22:34:39 +00:00
webmaster bd0881f882 cache-bust local assets; narrow the language facet to scores >= 3
Part 1 — asset versioning: every local <script src> and <link rel=stylesheet> on the
tracked pages now carries ?v=<sha256-of-the-file> (10 hex), computed at BUILD time from
the file's bytes by work/build-asset-versions.py — never hand-typed, and idempotent
(re-running with no file change rewrites nothing). This is what makes an edited
focus-rule.js / photo-upload.js a new URL instead of a stale cache hit.

Part 2 — language facet: a language now counts for the Language filter only when its
0-5 self-assessment is >= LANG_FACET_MIN (3 = 'works in it'). The constant sits beside
the 15-point rule in the page script and its mirror in build-directory.py. Nothing
stored changes; the card's languages chart still prints every language above 0, the
search box is unchanged, and the facet stays an exact membership test against the
card's value. A '1' language no longer offers the person under that language.
2026-11-03 21:26:58 +00:00
webmaster 17f171d3d9 feat(freelancer-answer): publish the private answer page; save via PATCH
Private utility page reached only by an emailed link (noindex, absent from
sitemap.xml/llms.txt, linked from nowhere). The freelancer reads the client's
question and sends an answer back through the odoo_experts_quote webhook.

- Save the answer with PATCH, not POST: the question record already exists
  (the client's submission created it), so the answer is a partial update of
  it. POST risks a second record.
- A method the route refuses (404/405/501) now fails honestly: "We could not
  save your answer", the answer is kept for a retry, and the receipt shows
  ONLY on a 2xx.
- Body shape (match in the n8n workflow): {"has":"<hash>","answer":"<text>"}.

Hash spelling: both ?has= and ?hash= are accepted on the way in; "has" is sent
outward. Question field name in the GET reply: unverifiable (no valid hash
exists), so the page names the fields it received instead of guessing.
2026-11-03 15:51:15 +00:00