fix(tide-display): load bitmap fonts at their own size, and document everything

[ChuckBuilds]

Tenth plugin in the README pass.

The font picker worked by coincidence

A .bdf exists at exactly one pixel size and FreeType rejects any other. So 4x6.bdf loaded — its native 6 happens to equal the default font_size: 6 — while 5x7.bdf silently fell back to the default face. Set font_size: 7 and the two would swap over.

before   66b4f94a29: ['4x6-font.ttf', '5x7.bdf']    <- 5x7 falling back
after    5 of 5 distinct

Same fix as #362 and #369. That's four plugins now; mqtt-notifications and youtube-stats still carry it — happy to sweep both in one PR if you'd prefer that to finding it plugin by plugin.

The README was a 70-line stub

None of the plugin's four screens were shown. It now covers all 18 settings with real screenshots of every screen, from recorded NOAA responses for Boston, and leads with the thing you actually have to do first — find your station ID.

Two behaviours that were undocumented and look like faults:

• A station with no live water-level sensor falls back to predictions rather than erroring. That's the common case, and it's logged at debug rather than as an error.
• Unit conversion is done by NOAA, not locally. The plugin asks the API for english or metric, so both are authoritative figures rather than one rounded from the other.

A mistake the tooling caught

The HTTP replay can now match on query parameters, not just the URL — NOAA puts three endpoints behind one URL and tells them apart by params, so without it every request got whichever fixture matched first.

That directly caught an error of mine. With the fixture ignoring units, metric mode rendered the imperial numbers with an "m" suffix — 9.7ft became "9.7m". It looked exactly like a missing unit conversion, and I was a step from writing it up as a bug. Reading the code showed the plugin correctly maps imperial→english and lets NOAA convert; the fault was entirely in my fixture. Recording both unit systems shows the real values (9.7ft / 3.0m).

One observation, not a diagnosis

On a 32-row panel the stats screen draws the range line and the percentage close enough to overlap at some values. It's in the README with images showing it, described as what it is rather than diagnosed — the chart and schedule screens are unaffected, and 64 rows clears it.

A note on render_preview.py

The plugin ships its own preview generator that re-implements the screens, with a header noting its palette "must match manager.py" — so its preview_*.png files can drift from real output without anything failing. The README images here come from running the actual plugin instead. I've left the script and its PNGs alone and just noted the difference; say the word if you'd like them removed, since they also ship to users' Pis on install.

Verification

• Audits: none dropped, 18/18 schema leaves documented, no broken anchors, every old section accounted for
• check_plugin.py --plugin tide-display — 32/32 PASS (8 sizes × 4 modes)
• render_docs_assets.py --plugin tide-display --check — images match
• manifest bumped 1.2.0 → 1.3.0

🤖 Generated with Claude Code

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Team

Run ID: 16fedf0a-c200-4193-80f7-4ecfd83e1970

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[codacy-production]

Not up to standards ⛔

🔴 Issues 1 medium

Alerts:
⚠ 1 issue (≤ 0 issues of at least minor severity)

Results:
1 new issue

Category	Results
Security	1 medium

View in Codacy

🟢 Metrics 14 complexity

Metric	Results
Complexity	14

View in Codacy

_{NEW Get contextual insights on your PRs based on Codacy's metrics, along with PR and Jira context, without leaving GitHub. Enable AI reviewer
TIP This summary will be updated as you push new changes.}

[ChuckBuilds]

Took you up on the offer and removed render_preview.py plus its seven preview_*.png files, since the README now has real renders that --check verifies.

The case for removing rather than keeping: the script drew its own mock versions of all four screens, with a header asking that its palette "must match manager.py". Nothing enforced that, so its committed previews could drift from actual output with no check to notice — and they shipped to every Pi on install, ~115KB the display never reads.

Worth contrasting with geochron, which has a render_preview.py of its own and does it right: it imports geochron_renderer, the same module manager.py draws with, so its previews cannot disagree with the plugin. If you want a standalone generator here again, that's the pattern — factor the drawing out and have both call it, rather than keeping a second copy of the palette. I've said as much in the commit message so the reasoning survives.

Nothing else in the repo referenced the removed files (checked across .md, .json, .py, .yml).

Separately, per your ask, I've started filing findings as issues rather than leaving them in PR descriptions — #371, #372, #373 here and ChuckBuilds/LEDMatrix#517 for the core font root cause.