[ChuckBuilds]

Stacked on #598 (base is its branch). Merge #598 first; GitHub then retargets this PR to main.

Why

Every scroll-strip build (each scroll-mode turn, and each Vegas content build) did avoidable setup work:

• A new GameRenderer per strip. Its constructor loads fonts and the element-style schema. In adaptive layout it also builds a LayoutContext, whose cache of fitted logos and text sizes was then thrown away.
• A fresh card-width probe on every read of the scroll settings. _default_game_card_width() ran on each read, building up to a dozen throwaway renderers to measure the score gap.

It is stacked on #598 because keeping the adaptive fit cache alive across builds would otherwise make #598's NFL/NCAA "logo:MIA" collision outlast a single build.

What

• ScrollDisplay._take_card_renderer / _keep_card_renderer:
  • The renderer is popped off the display for the build and put back after. That pop is a single dict operation, so two overlapping builds never share one.
  • A build that raises before putting it back just means the next build constructs a new one; nothing can get stuck.
  • It is reused only for the same config object, card width, panel height and logo cache. A config change rebuilds the whole scroll manager anyway (on_config_change).
• Rankings are set on every build (rankings_cache or {}), so a reused renderer cannot keep the previous build's ranks.
• _default_game_card_width() keeps its answer for the same config object and panel height. A failed probe is not kept.
• Memory stays bounded: the caches that now live across builds are already bounded LRUs (LayoutContext 512 fits / 64 images, the raw logo LRU 128).

Measured

Desktop, a 7-card strip at 192×48, 30 builds per process, run old/new/new/old:

layout	before (median)	after
classic	6.0 ms	3.4 ms
adaptive	7.2 ms	3.2 ms

Only 6 of the 14 test logos exist in the checkout, so real cards cost more. A Pi is roughly 10–20× slower in absolute terms.

Tests

test_scroll_renderer_reuse.py (new, standalone; needs LEDMATRIX_CORE) runs in both layouts and checks that:

• the second build constructs no renderer at all;
• a reused renderer draws a pixel-identical strip to a fresh one;
• rankings don't leak into a build that has none;
• a new config object gets a new renderer;
• a taken renderer is not handed to a second, overlapping build;
• it is handed back afterwards;
• a different card width gets a new renderer.

16 checks pass. On the old code it fails: the second build constructs 3 renderers.

Existing tests pass: test_core_scroll, test_vegas_elements, test_live_scroll_refresh, test_config_reload, test_vegas_content_rebuilds_on_change, test_vegas_score_clears_the_logos (19), and test_scroll_mode.py.

Guards: check_version_bump football-scoreboard (3.18.2 → 3.18.3), check_manifest_version_fields, check_manifests_ascii.

🤖 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: Advanced

Run ID: 246e64e5-302d-4989-94ce-17a214cdfb56

• Autopilot · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Autopilot is currently an internal CodeRabbit preview.

---

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]

Up to standards ✅

🟢 Issues 0 issues

Results:
0 new issues

View in Codacy

🟢 Metrics 31 complexity

Metric	Results
Complexity	31

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.}