feat: Add WeChat MiniGame build support and platform adapter - #1
Merged
Conversation
- Add platform adapter layer for Web and WeChat separation - Add WeChat-specific polyfills (fetch, wasm, fs) - Multi-entry build: index.js (Web), index.wechat.js (WeChat) - Add GitHub Actions workflow for npm publishing
- Add esengine_wxgame CMake target with WeChat-specific Emscripten flags (-sEXPORT_ES6=0, -sDYNAMIC_EXECUTION=0 for CommonJS compatibility) - Add scripts/build-wxgame.sh for automated WeChat engine builds - Add initRendererWithContext() to accept externally created WebGL context (avoids document.querySelector which doesn't exist in WeChat) - Update SDK createWebApp() to support glContextHandle option - Add Tauri commands for wxgame engine file access - Update WeChatBuilder to create WebGL context in JS and pass to WASM
esengine
added a commit
that referenced
this pull request
Jun 23, 2026
… P1.4d #1) SDF text was hardcoded to layer 0 → drawn under all UI quads (layer 1000+). Fix: a Text node in a UIRect is a UI render node — a PreUpdate system ensures it carries a UIRenderer (visualType None, not drawn) so the UI render-order pass assigns it a uiOrder; new ui_getRenderOrder binding exposes that order to TS, and the glyph batch uses layer = UI_BASE_LAYER + uiOrder. Pixel-verified: white text draws on top of a blue UIRenderer panel, centered + wrapped to the panel.
esengine
added a commit
that referenced
this pull request
Jun 23, 2026
…REARCH_GUI unification S4c) The legacy Transform-based ScrollView was a dual implementation (the rearch's #1 smell); the modern collection/ScrollContainer + ListView is the single scroll path. Per the clean-break directive: delete it outright (no migration) — ScrollViewPlugin.ts, ScrollView.ts, scrollview.test.ts, and all its plugin/ label/export references. One of the last two UIRect holders is gone. sdk tsc clean; 2297 tests green.
esengine
added a commit
that referenced
this pull request
Jun 23, 2026
…r<struct> (REARCH_GUI F5) Kills the audit's #1 worst component smell: StateVisuals' 8 hardcoded slotN* field quartets (32 flat fields) read by stringly-keyed reflection (record[`slot${i}Sprite`]). It's now a single variable-length `std::vector<VisualState>` looked up by name. EHT codegen — generalize to std::vector<CustomStruct>: - type_system: VisualState added to CUSTOM_STRUCTS (string+primitive members, vector-only so it skips the pointer cursor); is_struct_vector/vector_elem helpers; struct-vector forces a JS wrapper and marshals as emscripten::val. - embind: the wrapper field is an emscripten::val; from/toJS convert array↔vector element-by-element via the VisualState value_object (embind won't auto-convert a JS array to a registered std::vector, so a register_vector binding is not used). - field_utils: struct-vector defaults to [] and is editor-skip. Component + consumers: - C++ VisualState POD (flat r/g/b/a so it round-trips through the vector path); StateVisuals.states replaces the 32 slot fields. - TS state-visuals: states[] + a visualState(name, color, {sprite,scale}) builder; the apply system (behavior/systems) iterates states[] — no reflection. - button/dropdown build states[]; STATE_VISUALS_SLOT_COUNT removed. Verified: clean build -t web (ABI hash changed, expected), sdk tsc clean, full suite 2267 green incl. a new real-WASM embind round-trip test proving a vector<VisualState> (string + color + sprite + scale) survives the boundary; ui-node-flex pixel-verified (no render regression).
esengine
added a commit
that referenced
this pull request
Jun 29, 2026
…st (B4a) RC6 Batch B4 (compressed-texture producer — the #1 RC6 payoff: VRAM). The asset cook needs a build-time PNG/JPG -> KTX2 encoder; this lands the encoder + a tested API, decoupled from cook (no pipeline change yet, zero regression risk). - build-tools/basis/: the vendored prebuilt Basis Universal *encoder* wasm (basis_encoder.cjs + .wasm), built from the in-tree third_party/basis_universal via its reference emscripten project (non-threads wasm32 — Node-capable). Copied with a .cjs extension so Node loads it as CommonJS under this type:module repo, and committed (gitignore exception) so cooking needs no emsdk. README documents provenance + the rebuild command. - encoder.mjs: a small Promise API — encodePngToKtx2 / encodeToKtx2 (UASTC or ETC1S) / transcodeKtx2ToRgba. UASTC is emitted WITHOUT zstd supercompression to match the runtime transcoder (BASISD_SUPPORT_KTX2_ZSTD=0); same vendored library as the runtime transcoder, so encode and decode agree by construction. - ktx2-encode.test.ts: RGBA gradient -> KTX2 -> transcode round-trips at >40 dB PSNR (near-lossless UASTC); a real PNG encodes with matching dimensions; and the encoder is deterministic (identical bytes -> identical KTX2) — the prerequisite for content-addressing the .ktx2 in B3. Verified: 3 tests pass (desktop vitest). encode-time only; the runtime side (Batch C wiring) and cook integration (B4c) follow.
esengine
added a commit
that referenced
this pull request
Jun 29, 2026
…st (B4a) RC6 Batch B4 (compressed-texture producer — the #1 RC6 payoff: VRAM). The asset cook needs a build-time PNG/JPG -> KTX2 encoder; this lands the encoder + a tested API, decoupled from cook (no pipeline change yet, zero regression risk). - build-tools/basis/: the vendored prebuilt Basis Universal *encoder* wasm (basis_encoder.cjs + .wasm), built from the in-tree third_party/basis_universal via its reference emscripten project (non-threads wasm32 — Node-capable). Copied with a .cjs extension so Node loads it as CommonJS under this type:module repo, and committed (gitignore exception) so cooking needs no emsdk. README documents provenance + the rebuild command. - encoder.mjs: a small Promise API — encodePngToKtx2 / encodeToKtx2 (UASTC or ETC1S) / transcodeKtx2ToRgba. UASTC is emitted WITHOUT zstd supercompression to match the runtime transcoder (BASISD_SUPPORT_KTX2_ZSTD=0); same vendored library as the runtime transcoder, so encode and decode agree by construction. - ktx2-encode.test.ts: RGBA gradient -> KTX2 -> transcode round-trips at >40 dB PSNR (near-lossless UASTC); a real PNG encodes with matching dimensions; and the encoder is deterministic (identical bytes -> identical KTX2) — the prerequisite for content-addressing the .ktx2 in B3. Verified: 3 tests pass (desktop vitest). encode-time only; the runtime side (Batch C wiring) and cook integration (B4c) follow.
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
1 participant
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.
Summary
initRendererWithContext()to accept externally created WebGL context (avoidsdocument.querySelector)Changes
cmake/Emscripten.cmake- WeChat-specific build flags (-sEXPORT_ES6=0,-sDYNAMIC_EXECUTION=0)scripts/build-wxgame.sh- Automated build script for WeChat enginesrc/esengine/bindings/WebSDKEntry.cpp- AddinitRendererWithContext()sdk/src/app.ts- SupportglContextHandleoption increateWebApp()editor/src/builder/WeChatBuilder.ts- Generate game.js with JS-side WebGL context creationTest plan
document.querySelectorerrors