Journal¶
New findings land here first, then move to the page they belong on. Newest at the top.
2026-09-10 — the site went bilingual¶
Georgian stays primary; English lives at /en/. mkdocs-static-i18n with suffix
structure, fallback_to_default on so nothing 404s while pages are still being translated.
Custom domain fiber.qgis.ge — deliberately not fiberq., because half this site is
general FTTH material that has nothing to do with the plugin, and because fiberq. could
be mistaken for the official FiberQ site.
Order matters when adding a custom domain: DNS record first, docs/CNAME second. Do it
the other way round and GitHub Pages switches to the custom domain immediately, redirecting
the working github.io URL to a name that does not resolve yet.
2026-09-10 — the site was set up¶
MkDocs + Material. Zensical was not chosen: 0.0.60, Development Status :: 3 - Alpha,
still 0.0.x after ten months. Material for MkDocs is in maintenance mode but stable, and
ka is among its 69 UI languages. Zensical reads mkdocs.yml, so migrating later will be
easy.
Search limitation: lunr-languages does not support Georgian — no stemming, so ბოძი
will not match ბოძის. That is why the glossary is keyed on the English term.
A trap that nearly slipped through: MkDocs' default slugify strips non-ASCII, which
would have turned every Georgian heading into _1, _2 and broken all the anchor links.
Fixed with pymdownx.slugs.slugify.
2026-09-09 — the latent elements answer was already written¶
I asked the maintainer what List of latent elements meant — but the answer was in the
catalogue's own <extracomment>. The cause: when preparing the glossary I read the notes
in the UI-group contexts but not those in FiberQPlugin.
→ Latent element: a passive element sitting on a cable's path at a recorded distance along it, stored as data rather than drawn as a separate map feature.
Rule adopted: read all the notes first, then ask questions.
2026-09-09 — Color catalog is not symbology¶
In a QGIS context "Color catalog" reads naturally as a colour palette. It is in fact the fibre colour code — the TIA-598/IEC sequence identifying every tube and fibre in a cable.
To check: which standard does FiberQ use, TIA-598 or IEC? I have not looked at the code yet.
2026-09-09 — a counting error, and the correction¶
<message numerus="yes"> elements were not matching my count. The catalogue has 306
strings, not 293. Since the wrong numbers were already published in issue #43, a correction
was posted.
2026-09-09 — no language is actually translated¶
fiberq_fr.ts — 0/306. fiberq_sr.ts — 3/306 (test strings). The maintainer is Serbian and
has not translated his own language either.
Georgian will be the first real translation. That means the i18n pipeline has never been
exercised on a live catalogue, so the first one will surface problems. I found the first
already: _LANGUAGE_NAMES has no ka entry.
2026-09-09 — Object means building¶
The most consequential terminology finding. FiberQ's Object renders the legacy Serbian
objekat and means a building — confirmed by the layer it writes to (floors, basement
levels, street, house number).
Meanwhile Georgian QGIS translates feature as ობიექტი.
→ Feature → ობიექტი, Object → შენობა. Counter-intuitive, but correct.