Repository navigation
feat: phase 4 — matériaux, ville harmonisée et effets économiques - #10
Draft
mc-datanalytics wants to merge 5 commits into
Draft
mc-datanalytics wants to merge 5 commits into
mc-datanalytics wants to merge 5 commits into
Conversation
Complete the six remaining property models with distinct silhouettes, levels and LODs while preserving Roseraie and all earlier kits. Replace the old municipal pavilion and metro blocks, connect planted public walks, and improve integrated casino lettering. Add shared authored roughness/normal and ground contact/light fields, linear-light stylised surfaces, bounded economic-event particle batches, and quality/context lifecycle support. No game, network, UI, board IDs or economic rules changed. Team engagement branch is not integrated. Existing CI and session assertions retained; add pinned main before/after visual proof and separate ABBA software measurements. Executed locally: 255 Node tests, check, standalone build, harmony/public GLB and PNG exports and galleries. Source tree c610ae5 exactly matches the local validated tree. Native browser/visual validation pending CI; no real-device performance claim.
…ter native review The first real WebGL captures showed excessive regular wave bands and insufficient contrast behind the unchanged stage title. Add a bounded world-space sea haze and finer irregular ripples. Native game rendering worked, but the asset gallery exposed an unsupported re-export in the ESM-subset bundler; use an explicit import/export and add syntax validation of actual game/gallery bundles. 256 Node tests and standalone builds/galleries pass locally. Native proof and full CI are rerun; do not treat the previous failed gallery run as a complete validation.
First native ABBA diagnostics showed roughly 38–40% software-render overhead vs main. Skip the negligible specular/environment lobe on high-roughness materials and return the distant sea haze before ripple calculations. Glass/metal highlights, shared textures, property geometry, VFX, real reflection and functional assertions are retained. 256 local Node tests and standalone build pass. New native A/B and full CI must establish the actual cost; no speedup or hardware/mobile performance is presumed.
Native district regression found 18,949,206 bytes of geometry after camera/LOD traversal, exceeding its unchanged 18 MiB assertion. Bound Harmony by both 16 meshes and 4 MiB, evict only inactive least-recently-used models, and assert byte budget plus active-entry retention. All eight worst-case active high-detail variants fit in 3,616,392 bytes. No model/shader/UI/rule/network or existing test/workflow changed by this correction. Add the production/regeneration/budget/device protocol document and record previous bundle/background/memory failures and measured software overhead honestly. Executed 256 local Node tests, syntax/import checks and standalone build. Local and published tree e78d5d1 match. Full native regression and multiplayer are rerun on this revision; no main merge.
Logs from run 34158220245 show classic continuously progressing through all 147 UI commands, trace verification and lab import before CI marked timeout a few seconds beyond 22 minutes. Comp-60 likewise exported its complete session. This is a global runner-budget overrun, not proof of a passed CI or a network deadlock. Allow 25 minutes per full session step, 28 per job. Preserve viewport, real WebGL, scripts, scenarios, assertions, network convergence timeouts and validate job's original 20-minute bound. No render, geometry, gameplay, network or UI code changes in this commit. The previous 18 MiB geometry assertion is unchanged and fixed by byte-bounded cache in parent235702be. Document the failed runs and the distinction between functional validation and measured software overhead. Full CI reruns; no main merge.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Périmètre
Branche dédiée depuis le vrai
mainfusionnéb5f547c9. Ne pas fusionner sans validation du propriétaire. La branchefeat/engagement-rhythm-0-7reste indépendante.Validation à cet instant
Code initial :
b1a29df15b5b20e775e63060ecf8337bf4844660; arbre exactc610ae531ef15d8a0f2300d46b979afc53a4c6d6.Exécuté localement : 255 tests Node, contrôle de 88 fichiers JavaScript, build autonome, exports GLB/PNG de l'harmonisation et des équipements publics, galeries. La validation navigateur et l'examen des captures restent en cours ; le simple nombre de tests n'est pas une validation artistique.
Nouvelle recette en deux jobs séparés : vrai
mainépinglé pour l'avant, mêmes états légaux début/milieu/fin et mêmes caméras jour/crépuscule, rapprochements accessibles au joueur, studio ; mesures ABBA et cycles de ressources/VFX. Les suites de sessions multijoueurs d'origine sont conservées.Limites de méthode
Chromium/SwiftShader sur VM GitHub Actions : pas un test de performance sur téléphone physique ni TURN inter-réseaux. La carte de contact/lumière est une approximation stylisée, pas du SSAO ou de la GI ; les reflets de ciel sont analytiques. La référence artistique est une cible, pas un niveau de qualité déjà certifié. Les résultats finaux seront ajoutés après examen du moteur.