feat: add 17 items from PR #2866 + web articles + UX analysis (70→87)#41
Conversation
There was a problem hiding this comment.
Pull request overview
This PR updates the Qwen Code improvement comparison docs by adding newly identified Claude Code features (with source references) and expanding the terminal rendering/anti-flicker entry to reflect a deeper implementation than “fullscreen-only”.
Changes:
- Added new improvement entries in the main report index (including items 70–80) and updated item 54’s summary label/details.
- Expanded item 54 in the P2 detailed report from “Fullscreen Rendering” to a broader “terminal rendering optimization” description with implementation pointers.
- Added detailed write-ups for items 70–80 in the P2 detailed report (MCP reconnect, tool result size cap, max tokens escalation retry, ripgrep fallback/EAGAIN retry, magic docs, and input UX items).
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 4 comments.
| File | Description |
|---|---|
| docs/comparison/qwen-code-improvement-report.md | Adds new index rows for items 70–80 and updates the item 54 index entry to reflect expanded terminal rendering scope. |
| docs/comparison/qwen-code-improvement-report-p2.md | Expands item 54 with deeper rendering-engine details and appends detailed sections for items 70–80. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
New items discovered from 6 Claude Code architecture analysis articles: - wavespeed.ai, mindstudio.ai, zainhas.github.io, engineerscodex.com 71. Tool Result 大小限制 (maxResultSizeChars → 磁盘持久化) 72. Output Token 升级重试 (8K default → 64K on max_tokens truncation) 73. Ripgrep 三级回退 (System→Embedded→Builtin + EAGAIN -j1 retry) 74. MAGIC DOC 自更新文档 (# MAGIC DOC header → idle forked agent update) All verified in Claude Code source + confirmed missing in Qwen Code. 74 matrix items = 74 anchor links = 74 sections. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
New items focused on user experience improvements: 75. Ghost Text 输入补全 (P1) — inline dim suggestion for commands/paths/history 76. 目录/文件路径补全 (P2) — Tab complete paths + LRU cache 77. 上下文 Tips 系统 (P2) — tipRegistry with conditional display 78. 权限对话框文件预览 (P2) — file content + syntax highlight in approval 79. Token 使用实时警告 (P2) — live token warning + compression progress 80. 快捷键提示组件 (P2) — KeyboardShortcutHint with dynamic keybinding display Verified: Session Picker and Shell History already exist in Qwen Code (skipped). 80 anchors = 80 links = 80 matrix = 80 impacts, 0 broken. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
New items focused on interaction experience: 81. 终端完成通知 — iTerm2/Kitty/Ghostty OSC notifications + progress % 82. Spinner 工具名+计时 — "Bash(npm test) · 15s" instead of generic spinner 83. /rewind 检查点回退 — interactive checkpoint picker + file restore 84. /copy OSC 52 剪贴板 — SSH-friendly clipboard + temp file fallback 85. 首次运行引导向导 — multi-step onboarding (theme/auth/security/terminal) 86. /doctor 诊断工具 — environment health check + fix suggestions 87. 结构化 Diff 渲染 — Rust NAPI coloring + line gutter + syntax highlight Also added from previous round: 75. Ghost Text 输入补全 (P1) — inline dim command/path/history suggestion Verified: compact summary exists in Qwen Code (skipped), tool_group partial (skipped), @-mention Qwen-only (skipped). 87 anchors = 87 links = 87 matrix = 87 impacts, 0 broken. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
…SizeChars 1. Moved item 75 (Ghost Text P1) from P2 file to P0/P1 file 2. Updated main file link: p2.md#item-75 → p0-p1.md#item-75 3. Fixed item 71: maxResultSize → maxResultSizeChars (consistent naming) 4. Updated PR title to reflect actual 70→87 range Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
d2d165d to
923814b
Compare
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 3 out of 3 changed files in this pull request and generated 4 comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
1. Updated §三 navigation table: P0/P1 22→23, P2 37→53 (actual counts) 2. Fixed number format: 8,000 → 8_000, 4,911 → 4911, 64,000 → 64_000 (consistent with TypeScript literal style) 3. Double --- already clean (no action needed) Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 3 out of 3 changed files in this pull request and generated no new comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
P1 (1 item): - #27 Error recovery classification routing — truncation→continuation, overflow→compaction, transport→backoff, with per-category budgets P2 (2 items): - #40 drain-before-call — drain background notification queue before each LLM call so model sees latest results - #41 Post-compact identity re-injection — inject agent identity block when messages≤3 after compression (multi-agent scenario) All 3 items have full detail sections in their respective files with standard format (problem/solution/source/status/cost/comparison). Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Summary
5 new items from multiple sources:
Also expanded item 54 (terminal rendering) from Fullscreen-only to 8-layer anti-flicker.
All verified in Claude Code source + confirmed missing in Qwen Code.
🤖 Generated with Claude Code