Savings watermark failed on main #314

Activity

1. github-actions commented on Nov 2, 2026

   

   github-actionsBot

   on Nov 2, 2026 – with GitHub Actions

   Author

   More actions

   O job SIMUT savings watermark falhou: https://github.com/angeloINTJ/simut/actions/runs/41462746357 (falha simulada: não).
2. github-actions commented on Nov 3, 2026

   

   github-actionsBot

   on Nov 3, 2026 – with GitHub Actions

   Author

   More actions

   O job SIMUT savings watermark falhou: https://github.com/angeloINTJ/simut/actions/runs/41644496264 (falha simulada: não).
3. angeloINTJ

   added a commit that references this issue on Nov 4, 2026

   Merge pull request #316 from angeloINTJ/fix/savings-floor-tft-graph

   ...

   610d77d
4. angeloINTJ commented on Nov 4, 2026

   

   angeloINTJ

   on Nov 4, 2026

   Owner

   More actions

   The nightly job is green again on main: run 41813996703 (dispatched on 610d77d0, the merge of #316) prints check_savings: 16 recursos, nenhum abaixo do piso de economia., with tft_graph at 20135 B.

   What it was, measured on 2026-11-04 and written down in #316 (docs/analysis/ECONOMIA_DE_RECURSOS.md, section "O piso do tft_graph depois da caixa dos limites de alarme"): nothing of the graph leaked. #307 put a 68 B box in the middle of DisplayManager, and on the image without the graph that moved the members after it out of the Cortex-M0+'s cheap offsets (in loopCore1( ), 28 literal loads became 27 movs + lsls pairs): -238 B net in 65 functions. Moving the box to the end of the class would have brought the saving back, but grown pico_w_release by 8 B and pico2_w_release by 144 B, so the maintainer chose to lower the floor to 20,135 B.

   Still open, as its own change: tools/feature_costs.json (the configurator's estimate) predates #307, #308 and #311 and needs measure_savings.py --matrix.
5. angeloINTJ

   closed this as completedon Nov 4, 2026