fix: resolve the 12 remaining browser compatibility failures

[lanyue-llk]

原 372 项评测中的 12 个失败包含 10 项 Moli 实现缺陷、1 项实现与评测混合问题、1 项评测前提错误。本 PR 直接 target #640 的 codex/browser-eval-root-causes，当前基于 #640 头 45f8c72e，并已将自身提交真正 rebase 到该头部；两段历史均没有 merge commit。生产修改补齐共享协议、输入、网络终态、绘制链路与样式环境所有权；未按 case ID、页面结构或客户端名称添加分支。

• 由类型化域注册表统一 CDP 分发与 Schema；补齐受 --enable-automation 限制的真实启动参数查询、窗口 bounds 查询和局部更新语义；以现有 CDP 类型返回 computed style 所需字段。
• 区分文档加载完成与下载接管导航：后者发出 canceled net::ERR_ABORTED，使真实 Playwright 导航正确结束，文件仍完成下载并通过 checksum。
• 滚动接入既有 wheel 命中、事件取消和嵌套滚动管线。Inspector 高亮在独立绘制覆盖层解析实时节点几何，支持 DOM 兼容入口、隐藏/禁用和会话清理，不修改 DOM。
• 视觉模拟使用线性 RGB 滤镜并共享截图/录屏绘制；Inspector UI 在滤镜后合成。OS text scale 更新 Stylo CSS 环境变量及计算样式失效链，不缩放普通固定字号。

Stylo 的环境变量支持单独提交至 lexmount/stylo #1，其目标是 Moli 实际消费的 moli 分支。当前固定到个人 fork 的不可变 commit 3d44ed30808997861946a3f4b84a20318c4d451c；该提交基于 lexmount/stylo:moli@3c8bbb45，保留已有 content / counter() 等能力。这是样式引擎层修改，未在 JS/CSS 返回值表面替换结果。

评测实现另见 Lexbench #28，基于已有 #25：复用任务级 browser_automation 配置为双方启用 --enable-automation；视口先设置 780×600/DPR 1，保留原三条宽度断言并增加视觉视口断言。原报告冻结副本曾全局添加自动化标志，当前提交已用任务级配置替代。独立的 适配器可见性 PR #29 不参与这 12 项。

审查修复：Overlay.highlightNode 使用仓库现有的 CDP 类型解析完整已知参数；默认值标志与盒模型颜色一起正常高亮，需要额外绘制能力的选项（如 showInfo: true、网格配置和 selector）返回指明选项的协议错误。Overlay.enable 在有效 target 上直接完成；hideHighlight/disable 在无文档时可完成，在跨文档导航期间清理仍附着的旧页面，导航取消后高亮不会残留。

本轮审查修复：浏览器级媒体与文字缩放由 BrowserContext 持有默认值，并与 Page target 覆盖值统一合成；新 target、导航和页面会话解绑都读取同一有效状态。模糊视觉分块运行同一线性预乘高斯滤镜，使 4K 截图保持在瞬时内存预算内；删除没有生产消费者的 Overlay 布尔状态。Stylo Device setter 拒绝非有限或非正缩放，并在 Stylo 仓库直接测试 CSS 环境值与输入边界。

支持边界：滚动入口目前支持一次鼠标 wheel 手势；显式定时、触摸、fling、overscroll、重复等选项返回不支持。高亮支持盒模型颜色和矩形，不接受未实现的配置。窗口沿用现有虚拟窗口模型。视觉模拟不宣称不同栅格器逐像素等同，也不改变打印输出。

历史验证证据按提交区分：前序 f8006a62b 的二进制 SHA-256 为 5860b4969a83de031e33ef0354203cf7554ca41249da456e35e8ba2c6e2ee343，在原始 12 项的历史批次 pr731_review_12 中每项 3 次、Moli 36/36 通过；Chromium 沿用此前同评测合同的 36/36 对照。当前头 e692b1f7 已触发远端 CI；按本轮约定只执行了 cargo fmt --all 和 diff 检查，未在本地编译或运行测试。

审查图解与证据

完整验证报告：逐项根因、Chromium 对照与原始证据

以下为 ImageGen 生成的机制示意图；执行结果与原始证据见报告。

评测前提：自动化启动配置与显式视口

启动参数决定自动化接口是否可用；视口断言必须建立相同的 CSS 像素与 DPR 前提。

协议发现与类型化响应

域注册、分发与 Schema 使用同一数据源；响应字段满足真实客户端的类型化解析约定。

滚动事件与默认动作

滚动手势进入现有命中测试、wheel 事件取消与嵌套滚动链路。

绘制层次与实时几何

页面滤镜与 Inspector 覆盖层分别合成；高亮使用当前布局几何，不写入页面 DOM。

文字缩放：环境值、样式失效与重新计算

Moli 将缩放偏好传入 Stylo Device 并触发样式重算；Stylo 提供 env(preferred-text-scale)。只有引用该环境值的 CSS 参与缩放，普通固定字号保持原值。

下载接管与导航终态

附件下载接管导航后，导航结束与文件完成分别报告，避免客户端永远等待文档加载。

[lanyue-llk]

已对照 #640 的审查关注点检查协议边界、默认行为与生命周期，并核对本 PR 的改动及 CI。下面两处会影响新增 Inspector 接口的实际客户端使用；其余输入、下载和绘制路径未发现足以评论的确定性回归。

[lanyue-llk]

复查最新提交：上轮两条 Inspector 评论对应的单会话行为已修复。Rust 结构整体符合仓库既有风格；以下两处是可定位的功能边界，另有一处新增状态没有生产用途。

[lanyue-llk]

复查最新提交的跨目标状态、分块模糊、Inspector、输入及下载链路。上一轮三项修复在当前代码和新增测试覆盖的场景下成立；下面这处是浏览器范围状态的后续生命周期缺口。

[github-actions]

Sequential Navigation Soak A/B

✅ HEAD completed the 200-navigation resilience and memory observation.

One browser process, one target, and one CDP session navigate CSDN → SegmentFault → Huaban → example.com repeatedly. This run issued 200 Page.navigate commands per binary.

Common ancestor 45f8c72ef4 → HEAD 66a74a6754; benchmark order: base-first; workflow: success.

Session resilience

Metric	Base	HEAD
Attempted / planned	200 / 200	200 / 200
Direct observable pass	200 / 200	200 / 200
Failures / recovered	0 / 0	0 / 0
Recovery failures	0	0
Lifecycle order violations	0	0
Network order violations	0	0
Wall time	315.8 s	314.6 s

A public-page failure is reported separately from an unrecoverable session. The soak requires all navigation attempts, zero failed recovery, zero lifecycle/network ordering violations, and complete resource evidence.

Process-tree memory

Metric	Base	HEAD	Δ
Boundary samples	200	200	0
Boundary RSS samples	200	200	0
Boundary PSS samples	200	200	0
Periodic peak RSS	249.5 MiB	250.8 MiB	+1.3 MiB
Periodic peak PSS	247.8 MiB	247.3 MiB	-0.5 MiB
RSS · first 10 avg	174.2 MiB	171.8 MiB	-2.3 MiB
RSS · last 10 avg	163.7 MiB	162.2 MiB	-1.5 MiB
RSS · last − first window	-10.5 MiB	-9.7 MiB	+0.8 MiB
RSS · warm slope / 100 nav	+0.2 MiB	+0.7 MiB	+0.5 MiB
PSS · first 10 avg	172.0 MiB	168.5 MiB	-3.6 MiB
PSS · last 10 avg	160.3 MiB	158.5 MiB	-1.8 MiB
PSS · last − first window	-11.8 MiB	-10.0 MiB	+1.8 MiB
PSS · warm slope / 100 nav	-0.7 MiB	+0.6 MiB	+1.3 MiB
Peak file descriptors	64	63	-1
Peak threads	31	31	0

HEAD memory by 50-navigation quarter

HEAD range	Avg RSS	Final RSS	Peak RSS	Avg PSS	Final PSS	Peak FD
1–50	165.5 MiB	153.3 MiB	251.8 MiB	161.9 MiB	149.7 MiB	52
51–100	164.2 MiB	137.0 MiB	197.2 MiB	160.7 MiB	133.5 MiB	53
101–150	165.4 MiB	155.4 MiB	212.3 MiB	161.7 MiB	151.6 MiB	52
151–200	164.7 MiB	137.9 MiB	210.7 MiB	161.0 MiB	134.2 MiB	53

Raw reports and per-navigation boundary samples: workflow run and full sequential-navigation-soak-results artifact.

Public-site timing and memory are observational. Use the A/B deltas and trend shape as evidence, not as a deterministic performance threshold.

[github-actions]

Spider Bench A/B

✅ All benchmark browser service runs completed; results are informational.

Public HEAD: 179 / 240 rows; 37 / 48 sites produced rows; 11 unexpected empty sites.

✅ Deterministic fixture contract is clean.

Common ancestor 45f8c72ef475 → HEAD 66a74a675421; benchmark order: base-first; workflow: success.

Public 48-site run · informational

Metric	Base	HEAD	Δ
Run status	success	success	—
Extracted rows / contract	184 / 240	179 / 240	-5 (-2.72%)
Contract fill	76.67%	74.58%	-2.09 pp (-2.73%)
Sites with rows	38 / 48	37 / 48	-1 (-2.63%)
Sites meeting row contract	36 / 48	35 / 48	-1 (-2.78%)
Unexpected empty sites	10	11	+1 (+10.00%)
Partial-row sites	2	2	0 (0.00%)
Site outcomes	extracted: 36; HTTP error + rows: 2; empty: 7; HTTP error + empty: 1; snapshot error + empty: 2	extracted: 36; HTTP error + rows: 1; empty: 7; HTTP error + empty: 1; navigation error + empty: 1; snapshot error + empty: 2	—
Observed duration	56.18 s	86.01 s	+29.83 s (+53.10%)
Estimated CPU	53.60 s	58.19 s	+4.59 s (+8.56%)
Peak PSS	281.76 MiB	255.71 MiB	-26.05 MiB (-9.25%)
Peak RSS	283.80 MiB	259.18 MiB	-24.62 MiB (-8.68%)
Peak threads	27	26	-1 (-3.70%)
Mismatched sites	0	0	0
Timeout with items	0	0	0

Category rows	Base	HEAD	Δ
news	35 / 40	35 / 40	0
stocks	20 / 40	20 / 40	0
tech	35 / 40	35 / 40	0
sports	27 / 40	27 / 40	0
games	37 / 40	32 / 40	-5
life	30 / 40	30 / 40	0

Public-site content, timing, and memory are noisy. This single A/B run reports evidence only; site outcome counts explain missing rows without treating them as a deterministic regression.

CPU and memory timelines

These are bounded 40-point views of the same complete process-tree samples used by the HTML report. CPU uses 100% per occupied logical core; memory lines are RSS first and PSS second when complete PSS samples are available.

Base

xychart-beta
    title "Base CPU"
    x-axis "Elapsed seconds" [1.536, 3.038, 4.539, 6.541, 8.042, 9.544, 11.045, 12.547, 14.048, 15.549, 17.551, 19.051, 20.552, 22.054, 23.554, 25.056, 26.558, 28.06, 28.561, 30.064, 31.566, 33.067, 34.569, 36.071, 37.573, 39.075, 40.576, 41.077, 42.579, 44.08, 45.582, 47.084, 48.585, 50.085, 52.087, 53.589, 55.09, 55.591]
    y-axis "CPU percent" 0 --> 500
    line [125.9, 115.86, 123.87, 171.62, 175.89, 85.97, 175.53, 119.74, 75.99, 153.64, 2, 2, 113.75, 29.99, 39.97, 89.97, 137.91, 109.75, 101.82, 141.68, 107.97, 131.81, 101.86, 23.98, 65.96, 53.85, 0, 15.99, 115.93, 95.89, 4, 41.97, 116.07, 135.81, 79.89, 105.89, 119.92, 203.69]

Loading

xychart-beta
    title "Base memory: RSS then PSS"
    x-axis "Elapsed seconds" [0.034, 1.536, 3.038, 4.539, 6.541, 8.042, 9.544, 11.045, 12.547, 14.048, 15.549, 17.551, 19.051, 20.552, 22.054, 23.554, 25.056, 26.558, 28.06, 28.561, 30.064, 31.566, 33.067, 34.569, 36.071, 37.573, 39.075, 40.576, 41.077, 42.579, 44.08, 45.582, 47.084, 48.585, 50.085, 52.087, 53.589, 55.09, 55.591, 56.179]
    y-axis "Memory MiB" 0 --> 500
    line [11.98, 107.95, 117.13, 140.72, 148.77, 172.02, 156.21, 183.47, 167.13, 158.87, 163.54, 156.68, 137.95, 154.31, 146.24, 142.63, 149.32, 168.55, 283.8, 209.77, 203.46, 175.3, 167.55, 173.75, 148.38, 146.18, 147.84, 142.47, 143.25, 207.57, 182.16, 159.94, 162.34, 172.34, 164.51, 156.45, 147.86, 173.98, 198.52, 0]
    line [8.83, 104.25, 113.6, 137.06, 145.21, 168.31, 152.51, 179.85, 163.44, 155.31, 159.99, 153.13, 134.47, 150.83, 143.23, 139.23, 145.84, 165.13, 281.76, 205.7, 200.03, 170.4, 164.1, 170.39, 144.89, 142.7, 144.35, 138.98, 139.77, 204.25, 178.44, 156.44, 159.13, 168.03, 160.88, 152.8, 144.1, 170.43, 192.98, 0]

Loading

HEAD

xychart-beta
    title "HEAD CPU"
    x-axis "Elapsed seconds" [2.54, 5.041, 7.041, 7.542, 9.544, 12.047, 14.55, 16.553, 19.056, 21.56, 24.062, 26.564, 27.065, 28.568, 31.071, 33.574, 36.077, 38.08, 40.584, 43.086, 45.589, 48.092, 50.094, 52.598, 53.098, 55.1, 57.603, 59.606, 62.11, 64.613, 67.116, 69.619, 71.621, 74.124, 76.627, 79.13, 81.131, 83.634]
    y-axis "CPU percent" 0 --> 500
    line [7.99, 61.99, 151.99, 207.94, 165.92, 193.84, 78.01, 131.89, 119.83, 3.99, 43.94, 111.91, 107.81, 149.71, 149.65, 123.68, 37.98, 81.94, 119.89, 6, 23.98, 2, 2, 21.94, 0, 0, 19.99, 19.95, 0, 21.98, 0, 21.98, 2, 5.99, 11.99, 145.92, 150.2, 101.71]

Loading

xychart-beta
    title "HEAD memory: RSS then PSS"
    x-axis "Elapsed seconds" [0.035, 2.54, 5.041, 7.041, 7.542, 9.544, 12.047, 14.55, 16.553, 19.056, 21.56, 24.062, 26.564, 27.065, 28.568, 31.071, 33.574, 36.077, 38.08, 40.584, 43.086, 45.589, 48.092, 50.094, 52.598, 53.098, 55.1, 57.603, 59.606, 62.11, 64.613, 67.116, 69.619, 71.621, 74.124, 76.627, 79.13, 81.131, 83.634, 86.009]
    y-axis "Memory MiB" 0 --> 500
    line [12.15, 103.38, 121.41, 149.23, 138.58, 170.37, 177.92, 171.11, 187.14, 142.38, 147.55, 152.64, 213.75, 259.18, 197.29, 166.84, 174.65, 141.43, 148.13, 176.3, 239.88, 243.35, 244.69, 243.33, 243.25, 243.25, 243.21, 242.12, 242.13, 242.13, 242.12, 242.11, 242.1, 144.5, 159.07, 142.5, 173.19, 152.47, 151.5, 0]
    line [9.02, 99.93, 117.96, 145.77, 135.18, 166.88, 174.4, 168, 183.73, 138.9, 144.22, 149.15, 210.29, 255.71, 193.81, 163.43, 171.16, 137.95, 144.45, 173.06, 236.25, 239.79, 240.97, 239.61, 239.53, 239.53, 239.48, 238.39, 238.39, 238.39, 238.39, 238.39, 238.5, 140.95, 155.43, 138.85, 169.54, 148.82, 147.86, 0]

Loading

Deterministic fixture · required contract

The fixture exercises 15 routes: 8 are expected to emit rows and 7 intentionally exercise empty, loading, or timeout behavior.

Metric	Base	HEAD	Δ
Run status	success	success	—
Extracted rows / contract	40 / 40	40 / 40	0 (0.00%)
Contract fill	100.00%	100.00%	0.00 pp (0.00%)
Sites with rows	8 / 15	8 / 15	0 (0.00%)
Sites meeting row contract	15 / 15	15 / 15	0 (0.00%)
Unexpected empty sites	0	0	0
Partial-row sites	0	0	0
Site outcomes	extracted: 7; HTTP error + rows: 1; empty: 3; timeout + empty: 4	extracted: 7; HTTP error + rows: 1; empty: 3; timeout + empty: 4	—
Observed duration	19.87 s	19.87 s	-0.00 s (-0.01%)
Estimated CPU	0.39 s	0.39 s	-0.00 s (-0.25%)
Peak PSS	58.15 MiB	59.98 MiB	+1.83 MiB (+3.15%)
Peak RSS	61.28 MiB	63.36 MiB	+2.08 MiB (+3.40%)
Peak threads	18	18	0 (0.00%)
Mismatched sites	0	0	0
Timeout with items	0	0	0

Full HTML, JSON, CSV, logs, and page snapshots: workflow run and full spider-bench-results artifact.

The public-web A/B is informational. The exact HEAD fixture contract runs as a separate required CI check; timing and public-site content deltas remain non-blocking.

[github-actions]

CI Regression Report

Source CI run · source state at render: success · artifacts: 5/6

Check	Status	Signal
Release regression	✅	HEAD/base failures 0/0; raw binary +0.120642%
Frontend differential	✅	1,020/1,020 cases matched; 0 issues
Agent episodes	✅	8/8 Moli episodes passed; 0 failures
Runtime/CDP contracts	✅	26 contract cases; 0 failures
CDP smoke	✅	48/48 groups passed; 535 scenarios
WebMainBench	⚪	artifact unavailable or invalid

Release regression — ✅ HEAD/base failures 0/0; raw binary +0.120642%

Package and image size

Metric	Base	HEAD	Delta	Delta %
Raw binary	146.53 MiB	146.71 MiB	+181.02 KiB	+0.120642%
Stripped binary	100.58 MiB	100.74 MiB	+164.00 KiB	+0.159236%
gzip binary	50.16 MiB	50.25 MiB	+91.53 KiB	+0.178182%
Rootfs	153.44 MiB	153.61 MiB	+180.00 KiB	+0.114562%
gzip rootfs	58.94 MiB	59.04 MiB	+96.96 KiB	+0.160639%

Startup latency and PSS

Case	Base p50	HEAD p50	Delta	Base PSS	HEAD PSS	Delta
Serve ready	27.19 ms	27.16 ms	−0.03 ms	4.92 MiB	4.72 MiB	−212.00 KiB
CDP first page	49.72 ms	51.20 ms	+1.48 ms	4.03 MiB	4.13 MiB	+94.00 KiB
CDP warm page	12.18 ms	12.79 ms	+0.61 ms	49.55 MiB	49.01 MiB	−552.50 KiB
CLI about:blank	13.70 ms	14.26 ms	+0.55 ms	4.57 MiB	3.38 MiB	−1.19 MiB
CLI local JS	17.36 ms	16.93 ms	−0.43 ms	2.58 MiB	3.55 MiB	+990.00 KiB

Concurrency matrix

Concurrency	Base p50	HEAD p50	Delta	Base PSS	HEAD PSS	Delta	HEAD failures	Unstable B/H
1	16.29 ms	16.12 ms	−0.16 ms	683.00 KiB	673.00 KiB	−10.00 KiB	0	0/0
5	19.92 ms	20.44 ms	+0.52 ms	17.59 MiB	12.48 MiB	−5.11 MiB	0	0/0
10	19.27 ms	20.28 ms	+1.01 ms	26.79 MiB	14.79 MiB	−12.00 MiB	0	0/0
25	20.35 ms	19.80 ms	−0.56 ms	22.04 MiB	22.25 MiB	+221.25 KiB	0	0/0
100	19.94 ms	19.71 ms	−0.24 ms	17.56 MiB	16.33 MiB	−1.23 MiB	0	0/0

Frontend differential — ✅ 1,020/1,020 cases matched; 0 issues

Match	DOM mismatch	Diagnostic mismatch	Moli error	Reference error	Infrastructure error	Mismatched frames	Duration
1,020	0	0	0	0	0	0	6m 28.13s

Agent episodes · Moli vs Chromium — ✅ 8/8 Moli episodes passed; 0 failures

Fast contract	Episodes	Passed	Assertions	Ready p50	Episode p50
Moli	8	8	47/47	3.14 ms	25.68 ms
Chromium	8	8	47/47	244.27 ms	89.40 ms

Operation p50	Chromium	Moli	Delta	Delta %
navigate	39.78 ms	6.96 ms	−32.82 ms	−82.51%
observe	0.78 ms	0.85 ms	+0.08 ms	+9.64%
fill	3.23 ms	3.28 ms	+0.05 ms	+1.57%
click	4.41 ms	3.58 ms	−0.83 ms	−18.82%

Idle-resource episode	Peak RSS	Peak PSS	Average CPU	Peak CPU	Peak processes	Sampler
Moli	66.02 MiB	63.33 MiB	0.66%	10.00%	1	✅
Chromium	1281.89 MiB	332.30 MiB	1.81%	95.89%	13	⚠️ PSS partial

Runtime and CDP session contracts — ✅ 26 contract cases; 0 failures

Suite	Cases	Failures	Median case p50	Median case p95	Median peak PSS p50
Synthetic fetch	13	0	16.47 ms	17.21 ms	705.00 KiB
Long-lived CDP session	13	0	19.11 ms	19.29 ms	—

CDP smoke — ✅ 48/48 groups passed; 535 scenarios

Workers: 4 · cumulative group time: 143.47 s · failed groups: 0

Slowest group	Scenarios	Duration	Status
target-lifecycle	24	16.32 s	✅
multi-page	45	11.13 s	✅
chromium-cdp	56	10.68 s	✅
dom-input	27	9.50 s	✅
layout-screenshot	5	7.63 s	✅

WebMainBench · 545 pages — ⚪ artifact unavailable or invalid

Artifact unavailable or invalid. See the source CI run for infrastructure details.

All artifact fields are parsed by the trusted default-branch renderer; missing or invalid inputs remain visible as unavailable.