From 16a1139eab3e822dfc8ad2efba4dbc0a64bac7e9 Mon Sep 17 00:00:00 2001
From: funman300 <funman300@gmail.com>
Date: Mon, 6 Jul 2026 15:44:16 -0700
Subject: [PATCH] perf(web): size-focused wasm-release profile for the canvas
 build
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

canvas_bg.wasm shipped at 36.2 MB — plain release (opt-level 3, thin
LTO) piped through wasm-opt -O2. Download size, not throughput, is the
binding constraint for the browser canvas, so solitaire_web now builds
with a dedicated wasm-release profile: opt-level "s", fat LTO, one
codegen unit. Local result: 23.2 MB after the unchanged wasm-opt -O2
pass — 35.9% smaller.

The binaryen pass stays at -O2 (the -Oz grey-screen miscompile note in
build_wasm.sh still applies). profile.strip is deliberately NOT set: on
wasm it also removes the target_features custom section, which makes
wasm-opt reject the module ('all used features should be allowed' on
trunc_sat).

Verified with the new artifact: all 5 play_canvas e2e specs pass (debug
bridge, draw3 param, apply/undo, replay diagnostics, 40-seed autoplay
invariants). Headless pixel verification is inconclusive in this
environment — wgpu cannot create a usable adapter locally and panics
identically on the OLD artifact too — so visual parity should be
confirmed against production after the next deploy.

pkg/ artifacts are intentionally not committed: the web-wasm-rebuild
workflow is the single source of truth and will regenerate them with
this profile on merge.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---
 Cargo.toml    | 16 ++++++++++++++++
 build_wasm.sh |  6 ++++--
 2 files changed, 20 insertions(+), 2 deletions(-)

diff --git a/Cargo.toml b/Cargo.toml
index 3ddcf9e..2a7f490 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -156,3 +156,19 @@ opt-level = 3
 [profile.release]
 opt-level = 3
 lto = "thin"
+
+# Size-focused profile for the browser canvas build (solitaire_web →
+# canvas_bg.wasm). Download size is the constraint on the web, not peak
+# throughput: fat LTO + one codegen unit + opt-level "s" cut the Bevy wasm
+# bundle substantially versus plain release. This is rustc-side sizing only —
+# the binaryen pass in build_wasm.sh stays at wasm-opt -O2 because -Oz has
+# miscompiled Bevy's render pipeline before (grey screen on first load).
+[profile.wasm-release]
+inherits = "release"
+opt-level = "s"
+lto = "fat"
+codegen-units = 1
+# No `strip`: on wasm it also removes the target_features custom section,
+# which makes wasm-opt reject the module ("all used features should be
+# allowed" on trunc_sat). wasm-opt drops the name section in its output
+# anyway, so strip buys nothing here.
diff --git a/build_wasm.sh b/build_wasm.sh
index 995d17d..cff37ab 100755
--- a/build_wasm.sh
+++ b/build_wasm.sh
@@ -67,7 +67,9 @@ if ! command -v wasm-bindgen &> /dev/null; then
 fi
 
 echo "Building solitaire_web (Bevy WASM app)..."
-cargo build --release --target wasm32-unknown-unknown -p solitaire_web
+# wasm-release is the size-focused profile (fat LTO, CGU=1, opt-level "s") —
+# see [profile.wasm-release] in Cargo.toml. Download size is the constraint.
+cargo build --profile wasm-release --target wasm32-unknown-unknown -p solitaire_web
 
 echo "Running wasm-bindgen for solitaire_web..."
 wasm-bindgen \
@@ -75,7 +77,7 @@ wasm-bindgen \
     --out-name canvas \
     --target web \
     --no-typescript \
-    "$REPO_ROOT/target/wasm32-unknown-unknown/release/solitaire_web.wasm"
+    "$REPO_ROOT/target/wasm32-unknown-unknown/wasm-release/solitaire_web.wasm"
 
 # Optional size optimisation — Bevy bundles are large (~5-15 MB uncompressed).
 # wasm-opt passes are skipped silently when the tool is not installed.