fix(cli): default to virtualized terminal history - #5738
Conversation
|
E2E / TUI verification report for this PR:
Additional local checks: Tested locally on macOS. Windows/Linux terminal combinations were not manually tested. |
wenshao
left a comment
wenshao
left a comment
There was a problem hiding this comment.
No review findings. Downgraded from Approve to Comment: CI failing: Test (windows-latest, Node 22.x).
CI note: the failing Windows job is in packages/core's gitWorktreeService.symlinks.integ.test.ts (git init -q -b main), outside this PR's changed files.
— GPT-5 Codex via Qwen Code /review
Verification Report — default to virtualized terminal historyBuilt and exercised the real CLI from the PR head, with a focus on the one thing a default-flip can silently get wrong: leaving a runtime gate behind. Result: correct, complete, and well-tested. The only open item is a product/UX judgement call (alt‑screen now on by default), not a defect. Environment: PR head Scope checkThe PR is a focused 10‑file, single‑commit change that flips 1. Gate‑completeness audit ✅ (avoids the schema‑only trap)A schema This PR does not fall into that trap. There are exactly two runtime reads of
2. Real tmux E2E (A/B) ✅ — the headlineTwo server‑side probes were used because they independently exercise the two different gates:
The third row is a dist‑level swap ( 3. UX consequence reviewers should weigh
|
| Gate | 控制 | 状态 |
|---|---|---|
gemini.tsx:367(useVP) |
render(..., { alternateScreen: useVP }) |
?? false → ?? true ✅ |
AppContainer.tsx:957(useTerminalBuffer) |
VP 渲染路径 + refreshStatic;经 uiState.useTerminalBuffer 传给 MainContent/ScrollableList |
?? false → ?? true ✅ |
MainContent.tsx 通过 uiState 继承该值(没有独立读取)。settingsSchema.ts 和 vscode‑ide 的 JSON 默认值也一并翻转,所以设置对话框与运行时保持一致。serve/routes/workspace-settings.ts 那处只是一个 TUI‑only 分类集合(default 字段自动从 schema 派生)。ACP / serve / 非交互路径都不读这个设置 —— 改动仅限 TUI。由于用的是 ??,未设置的 key 会正确落到新默认值,与 merge 是否 seed 无关。
2. 真实 tmux E2E(A/B)✅ —— 核心证据
用了两个 server 端探针,因为它们分别独立地命中两个不同的 gate:#{alternate_on} ← gemini.tsx,#{mouse_any_flag} ← AppContainer → ScrollableList。同一个构建,两个隔离的 $HOME,两个 arm 都干净地到达聊天输入框(无 auth 对话框)。
| 构建 / 配置 | #{alternate_on} |
#{mouse_any_flag} |
含义 |
|---|---|---|---|
| PR + 设置未设(新默认) | 1 | 1 | 默认开启 VP |
PR + ui.useTerminalBuffer: false(opt‑out) |
0 | 0 | 旧路径保留 |
| base(在 dist 里还原 PR 改动)+ 未设 | 0 | 0 | 证明这 2 行改动是唯一原因 |
第三行是 dist 级替换(把两行编译产物 ?? true → ?? false,$HOME 同样未设),从而把 PR 改动隔离成唯一变量。两个 gate 都验证通过,逃生开关也验证通过。
3. Reviewer 需要权衡的 UX 影响 ⚠️ (非 bug)
新默认会进入 alternate screen。已具体演示:在启动前打印的 host marker,在 qwen 运行期间被隐藏(alternate_on=1),/quit 之后被还原,且 qwen 界面消失:
运行期间: alternate_on=1 | qwen banner 可见 | host marker 被隐藏
/quit 之后: alternate_on=0 | qwen banner 消失 | host marker 还原
也就是说退出时整个 qwen 会话会从宿主终端 scrollback 中抹掉(标准的 vim/less 式 alt‑screen 行为)。这是为消除闪屏而做的取舍,文档里的 opt‑out(ui.useTerminalBuffer: false)有效 —— 但对那些退出后还要在终端往回滚、或依赖原生 click‑drag 选择文本的用户来说是个实质变化。建议产品侧明确签字确认。
4. 单元测试 ✅
- PR 改动的 3 个文件:147 / 147 通过(与 PR 描述一致)。
- 对 6 个受 default 翻转影响的测试套件做反向审计扫描(
MainContent、App、DefaultAppLayout、useResizeSettleRepaint、terminalRedrawOptimizer、InputPrompt):221 / 221 通过。PR 正确地在共享的AppContainerfixture 里钉死useTerminalBuffer: false,使依赖旧路径的测试(如 Terminal resize during streaming leaves fragmented content at wrong widths in scrollback #4891 的 resize‑clearTerminal 契约)继续有效。未发现连带破坏。
5. CI 状态
Test (ubuntu) 和 Test (macos) 通过。Test (windows) 失败,但失败项是 gitWorktreeService.symlinks.integ.test.ts(15 中 1 个)外加 exit code 127(shell command‑not‑found)—— 这是 Windows symlink/工具链问题,而该文件本 PR 完全没动(merge‑base 与 PR head 逐字节相同)。不是本 PR 引入的回归。
结论
从正确性角度,本 PR 可以合并:gate 审计干净,两个 runtime fallback 与显示默认值一致翻转,Linux/macOS 测试全绿,真实终端 A/B 同时验证了新默认与 opt‑out。剩下的只是"是否默认进入 alternate screen"的产品决定(见 §3)。Windows CI 的红是既有/无关问题。
Verified by building the PR head and running it under tmux on Linux/Node 22. Probes: #{alternate_on} (gemini.tsx gate) and #{mouse_any_flag} (AppContainer→ScrollableList gate), plus a dist‑level base A/B isolating the PR delta.
|
Follow-up on the remaining UX point: Defaulting The compatibility path is preserved: users who prefer native host scrollback can set I do not plan further code changes for this unless maintainers prefer a more conservative rollout strategy than default-on. CI note: the current Windows red is in |
hi, thank you for the PR. But I think it's not time to change the default setting of
If you glad with this, you can help to test and welcome pr with this. |
doudouOUC
left a comment
doudouOUC
left a comment
There was a problem hiding this comment.
[Suggestion] docs/design/virtual-viewport/README.md:326 — The paragraph after the PR-sequence table still reads "V.3 (integration tests) is the remaining critical-path item before flipping the default." However, V.3 remains "pending" in the table while this PR flips the default. Future readers will conclude the default was flipped prematurely or that V.3 was silently dropped. Suggest updating to reflect the actual decision, e.g.: "V.3 (integration tests) remains desirable for long-session regression coverage but is no longer a gating prerequisite for the default flip."
— qwen3.7-max via Qwen Code /review
| // re-reading `mergedHistory` / `allVirtualItems` on whatever state | ||
| // change triggered refreshStatic (Ctrl+O, model change, etc.). | ||
| const useTerminalBuffer = settings.merged.ui?.useTerminalBuffer ?? false; | ||
| const useTerminalBuffer = settings.merged.ui?.useTerminalBuffer ?? true; |
There was a problem hiding this comment.
[Suggestion] Accessibility regression for screen-reader users. Flipping the default to true silently routes screen-reader users into the virtualized-viewport path, which drops the append-only <Static> rendering that screen-reader mode is built around.
Chain: with ui.useTerminalBuffer unset, this is now true → uiState.useTerminalBuffer is true → MainContent takes the if (useVirtualScroll) branch (MainContent.tsx:596) and renders <ScrollableList> with no <Static> (the legacy <Static> branch is at :626). Screen-reader users reach this same MainContent — App.tsx routes them to ScreenReaderAppLayout, which itself renders <MainContent />. Ink's screen-reader render path emits committed history to the terminal only via staticOutput (a <Static> node); with no <Static>, history is no longer presented append-only. Separately, gemini.tsx:367 now passes alternateScreen: true for screen-reader + TTY (Ink's resolveAlternateScreenOption doesn't exclude screen-reader mode), entering the alternate screen that assistive tech / native scrollback rely on.
Before this PR (default false) screen-reader users got the <Static> path. Suggest gating the default on screen-reader mode here (and mirror it at gemini.tsx:367 for alternateScreen):
| const useTerminalBuffer = settings.merged.ui?.useTerminalBuffer ?? true; | |
| const useTerminalBuffer = | |
| (settings.merged.ui?.useTerminalBuffer ?? true) && !config.getScreenReader(); |
No test currently covers the screen-reader + unset-useTerminalBuffer combination.
— claude-opus-4-8[1m] via Qwen Code /qreview
There was a problem hiding this comment.
Fixed in 90ea976.
Screen-reader mode now stays out of the VP and alternate-screen path in both runtime gates: AppContainer keeps uiState.useTerminalBuffer false and startInteractiveUI passes alternateScreen false when config.getScreenReader() is true. Added regression coverage for screen reader mode with unset useTerminalBuffer in AppContainer and gemini.
ee5d0ea to
90ea976
Compare
|
Updated PR head to 90ea976 after #5751 landed. Handled review items:
Verification:
Local build note:
|
Build failure root cause — it's a
|
wenshao
left a comment
wenshao
left a comment
There was a problem hiding this comment.
[Suggestion] packages/cli/index.ts (not in diff) — The uncaughtException handler calls process.exit(1) without runExitCleanup(), so instance.unmount() (which writes exitAlternateScreen) never runs. Pre-existing bug, but now that VP is the default and all users enter alt-screen, any crash leaves the terminal stuck in alt-screen with no visible prompt (user must run reset or close the tab).
Fix: add a process.on('exit') handler that unconditionally writes \x1b[?1049l as a safety net, or call runExitCleanup() in the uncaughtException handler.
CI note: 3 Test jobs failing (ubuntu, macos, windows) — pre-existing, not caused by this PR.
— qwen3.7-max via Qwen Code /review
|
|
||
| const useVP = settings.merged.ui?.useTerminalBuffer ?? false; | ||
| const useVP = | ||
| (settings.merged.ui?.useTerminalBuffer ?? true) && |
There was a problem hiding this comment.
[Suggestion] TOCTOU between one-shot alternateScreen and per-render useTerminalBuffer: useVP is computed once here and passed to ink's render(), locking the terminal into alt-screen for the entire session. But AppContainer.tsx:958 recomputes useTerminalBuffer on every render from live settings.merged. If a user toggles useTerminalBuffer off mid-session (the schema declares requiresRestart: false), the React tree switches to the legacy <Static> path while ink stays in alt-screen — history is silently written to the alt-buffer and lost on exit.
Pre-existing, but the default flip makes this the common path. Consider either locking useTerminalBuffer at startup (matching alternateScreen's lifetime), or setting requiresRestart: true so the settings dialog prompts a restart.
— qwen3.7-max via Qwen Code /review
There was a problem hiding this comment.
Fixed in 8f5eb98.
The VP decision is now startup-scoped in AppContainer, matching Ink alternateScreen lifetime. ui.useTerminalBuffer is also marked requiresRestart, so SettingsDialog no longer presents this as a live-safe toggle.
| // change triggered refreshStatic (Ctrl+O, model change, etc.). | ||
| const useTerminalBuffer = settings.merged.ui?.useTerminalBuffer ?? false; | ||
| const useTerminalBuffer = | ||
| (settings.merged.ui?.useTerminalBuffer ?? true) && |
There was a problem hiding this comment.
[Suggestion] The VP-mode guard (settings.merged.ui?.useTerminalBuffer ?? true) && !config.getScreenReader() is duplicated verbatim in gemini.tsx:368. These two sites control different but tightly coupled decisions (ink's alternateScreen vs React rendering path) and must always agree.
Consider extracting a shared helper:
// e.g., packages/cli/src/ui/utils/terminalBuffer.ts
export function shouldUseVirtualViewport(
settings: LoadedSettings,
config: Config,
): boolean {
return (settings.merged.ui?.useTerminalBuffer ?? true) && !config.getScreenReader();
}— qwen3.7-max via Qwen Code /review
There was a problem hiding this comment.
Fixed in 8f5eb98.
Both runtime gates now use a shared shouldUseVirtualViewport helper, so the alternate-screen decision and React VP path stay coupled.
| expect(options).toMatchObject({ alternateScreen: false }); | ||
| }); | ||
|
|
||
| it('should not use alternate screen in screen reader mode when VP mode is unset', async () => { |
There was a problem hiding this comment.
[Suggestion] Test gap: the screen-reader guard is tested when useTerminalBuffer is unset (defaults to true), but not when it's explicitly true. Add a test with useTerminalBuffer: true + getScreenReader(): true to cover the true && !true branch:
it('should not use alternate screen in screen reader mode even when VP mode is explicitly enabled', async () => {
// useTerminalBuffer: true, getScreenReader: () => true
// expect alternateScreen: false
});— qwen3.7-max via Qwen Code /review
There was a problem hiding this comment.
Covered in 8f5eb98.
Added the explicit useTerminalBuffer: true + screen-reader case for startInteractiveUI, asserting alternateScreen remains false.
90ea976 to
8f5eb98
Compare
|
Updated PR head to 8f5eb98 after reviewing the latest comments and CI failure. What changed:
Verification:
|
doudouOUC
left a comment
doudouOUC
left a comment
There was a problem hiding this comment.
No review findings. LGTM! ✅
All prior suggestions (duplicated VP formula, missing screen-reader test coverage) have been addressed — shared shouldUseVirtualViewport() helper, session-locked VP mode via useState, alternate-screen exit handler, and requiresRestart: true are clean improvements.
— qwen3.7-max via Qwen Code /review
doudouOUC
left a comment
doudouOUC
left a comment
There was a problem hiding this comment.
No new code changes since last review (only a merge from main). Re-approving on the updated commit. LGTM! ✅
— qwen3.7-max via Qwen Code /review
|
@qwen-code /resolve |
|
Qwen Code did not run conflict resolution for this request. PR #5738 comes from ZevGit/qwen-code; this first version only pushes same-repository branches. |
# Conflicts: # packages/cli/src/gemini.tsx
|
Synced this branch with upstream During conflict review, I preserved the current upstream TUI behavior (including transcript streaming context, Kitty protocol handling, mouse selection/click behavior, and startup cleanup) while keeping the VP decision centralized and frozen for the Ink alternate-screen lifetime. The latest upstream also added a live-agent panel row cap; I updated its keyboard-selection path to use the effective Verification on the merged tree:
This PR has already gone through more than five review rounds, so I am freezing scope to Critical fixes only. The existing non-blocking suggestions (the configuration reference table default and an optional debug breadcrumb/additional narrow test) are intentionally deferred rather than expanding this PR again. I will monitor the refreshed CI and address only failures attributable to this change. |
qwen-code-ci-bot
left a comment
qwen-code-ci-bot
left a comment
There was a problem hiding this comment.
Reviewed. Suggestions are inline.
— qwen3.7-max via Qwen Code /review
| ): boolean { | ||
| // The settings loader does not apply schema defaults, so keep this fallback | ||
| // in sync with settingsSchema.ts's default for ui.useTerminalBuffer. | ||
| return terminalInteractive && (useTerminalBuffer ?? true) && !screenReader; |
There was a problem hiding this comment.
[Suggestion] The ?? true fallback is a silently-coupled duplicate of settingsSchema.ts's useTerminalBuffer default — Failure scenario: a future maintainer changes the schema default to false (e.g., to stage a rollback) but misses this line; users with no explicit setting still enter VP/alt-screen while the settings dialog shows "Default: false". The test assertion in terminal-buffer.test.ts guards against drift, but the coupling is documentary (comment) rather than structural.
| return terminalInteractive && (useTerminalBuffer ?? true) && !screenReader; | |
| const defaultVP = getSettingsSchema().ui.properties.useTerminalBuffer.default as boolean; | |
| return terminalInteractive && (useTerminalBuffer ?? defaultVP) && !screenReader; |
— qwen3.7-max via Qwen Code /review
| function isCiEnvironment(env: TerminalEnvironment): boolean { | ||
| return Object.keys(env).some( | ||
| (key) => isCiEnvKey(key) && isActiveCiValue(env[key]), | ||
| ); | ||
| } |
There was a problem hiding this comment.
[Suggestion] isCiEnvironment only detects CI via CI, CONTINUOUS_INTEGRATION, and CI_*-prefixed env vars, missing several major CI systems that may not set CI=true (Jenkins: JENKINS_URL/BUILD_NUMBER, TeamCity: TEAMCITY_VERSION, Azure DevOps: TF_BUILD, AWS CodeBuild: CODEBUILD_BUILD_ID) — Failure scenario: a Jenkins pipeline runs qwen inside docker run -it with TTY but no CI env var; isInteractiveTerminal() returns true, Ink enters alt-screen, and the CI log capture misses the entire conversation.
Consider adding vendor-specific CI env var checks:
const VENDOR_CI_KEYS = [
'JENKINS_URL', 'TEAMCITY_VERSION', 'TF_BUILD',
'CODEBUILD_BUILD_ID', 'BITBUCKET_BUILD_NUMBER',
];— qwen3.7-max via Qwen Code /review
qwen-code-ci-bot
left a comment
qwen-code-ci-bot
left a comment
There was a problem hiding this comment.
— qwen3.7-max via Qwen Code /review
Local verification report — default to virtualized terminal historyBuilt both sides from source and drove the real CLI under a real pty on macOS, focusing on what a default-flip can silently get wrong: a runtime gate left behind, a guard that only exists in the schema, or an interaction that regresses for everyone instead of the opt-in minority. Result: the implementation is correct and complete. All twelve behavioural cells match intent, the two interaction bugs raised in this thread in June are fixed on the current head, and the 2-gate delta is provably the sole cause of the change. Two items below are worth a conscious call before merge — neither is a defect in this diff. Environment: PR head 1 · Gate audit — no schema-only trapA schema
Coupling the two gates through one prop is the right call — 2 · Real A/B matrix — twelve cells, two independent gatesProbes are read straight off the wire:
Every guard is load-bearing, not decorative. Two worth calling out:
The 3 · The two blockers raised in this thread — re-testedContent streamed from a local mock OpenAI server; mouse input delivered as genuine SGR reports into the pty, exactly as a terminal emulator sends them.
Both were listed as reasons to hold the default flip. On this head they behave correctly. 4 · Tests — 276/276 green, and they hold the behaviour in placeAll 9 PR-touched test files pass (276 tests). To check the tests are actually load-bearing rather than merely green, I broke one behaviour at a time in the source and re-ran them:
Finding 1 (test quality, non-blocking). The lock itself is fine — I verified it in the real app by flipping 5 · Exit hygiene with alternate screen on by defaultEach run prints host-shell scrollback, launches qwen, terminates it a different way, then prints more shell output. Oracle is xterm's own buffer type — the same state tmux exposes as
Good news first: Finding 2 (consequence of the default, pre-existing, non-blocking). This is not a regression in this diff — the base build reproduces it identically once It is also not a one-liner — I tested two candidate fixes and both failed:
The reason ink's own teardown never runs is VerdictFrom a correctness standpoint this is ready to merge: the gate audit is clean, both display and runtime defaults flip consistently, the escape hatch and all four guards (screen-reader, CI, dumb terminal, non-TTY stdout) are verified on a real terminal, tests are green and — with one exception — mutation-proof, and the interaction bugs that motivated holding this back are fixed. The two open items are (1) a vacuous session-lock test worth tightening, and (2) a pre-existing signal-teardown gap that this flip makes universal. Neither blocks the diff; (2) is the one I'd want tracked before this reaches a release. 中文版（点击展开）本地验证报告 — 默认开启虚拟化历史视图从源码构建了 PR 与 base 两侧，并在 macOS 真实 pty 中驱动真实 CLI，重点检查 default 翻转最容易悄悄出错的地方：是否漏掉某个 runtime gate、是否只改了 schema、以及是否让某个交互从"少数 opt-in 用户受影响"变成"所有人受影响"。 结论：实现正确且完整。 12 个行为单元全部符合预期；6 月在本 thread 中提出的两个交互 bug 在当前 head 上已修复；并且已证明这 2 个 gate 的改动是行为变化的唯一原因。下面两项建议在合并前明确确认一次 —— 两者都不是本 diff 的缺陷。 环境： PR head 1 · Gate 审计 —— 没有掉进 schema-only 陷阱schema 里的
用一个 prop 把两个 gate 耦合起来是正确做法： 2 · 真实 A/B 矩阵 —— 12 个单元，两个独立 gate探针直接读取 wire 上的字节： 其中两点值得特别说明：
3 · 本 thread 中提出的两个 blocker —— 重新测试内容由本地 mock OpenAI 服务流式返回；鼠标事件以真实 SGR 报文写入 pty，与终端模拟器发送的完全一致。
这两点此前被列为暂缓翻转默认值的理由，在当前 head 上均表现正常。 4 · 测试 —— 276/276 通过，且大部分确实"钉住"了行为PR 改动的 9 个测试文件全部通过（276 个用例）。为了确认这些测试是真的有约束力而不只是"碰巧绿"，我逐个破坏源码中的行为并重跑：
问题 1（测试质量，不阻塞）。 锁本身没有问题 —— 我在真实应用里验证过：运行中把 5 · 默认进入 alternate screen 后的退出清理每次运行都会先打印宿主 shell 的 scrollback，再启动 qwen，用不同方式终止，然后再打印 shell 输出。判据是 xterm 自身的 buffer type —— 即 tmux 的 先说好消息： 问题 2（默认值带来的后果，且为既有问题，不阻塞）。 这不是本 diff 引入的回归 —— base 构建只要设置 同时它不是一行能修好的 —— 我测了两个候选修法，都无效：
ink 自身的 teardown 之所以不执行，原因在 结论从正确性角度，本 PR 可以合并：gate 审计干净，显示默认值与运行时默认值一致翻转，逃生开关以及四个 guard（screen-reader、CI、dumb terminal、非 TTY stdout）都在真实终端上验证通过，测试全绿且除一处外都能通过变异检验，此前促使暂缓的交互 bug 也已修复。 两个待办项：(1) 一条空转的 session-lock 测试值得收紧；(2) 一个既有的信号清理缺口被这次翻转放大到所有用户。两者都不阻塞本 diff；其中 (2) 建议在发版前有记录地跟踪。 Verified by building both merge-base and PR head in isolated worktrees (independent |
|
Follow-up on the terminal-teardown findings from the latest real-PTY verification:
I kept both out of this already heavily-reviewed PR so the distinct root causes can be handled and tested cleanly. CI for the current head is green, and the full VP behavior matrix remains verified. @chiga0 Could you confirm whether #7779 should be fixed before merging this default-on change, or whether both follow-ups should be tracked for the V1 stabilization window? I will keep #5738 scope frozen unless a maintainer classifies one as a merge blocker. |
Both #7779 (Kitty keyboard cleanup ordering) and #7781 (SIGTERM/SIGHUP teardown gap) are pre-existing issues on The PR itself has been thoroughly reviewed and verified (multiple real-PTY A/B tests, maintainer sign-off). The only remaining hard blocker is the merge conflict — please rebase onto the latest |
|
Synced the branch to |
|
@qwen-code /takeover |
|
@qwen-code-ci-bot /review |
|
🚫 Takeover not engaged: fork takeover requires the PR author to hold write access on this repository (author 中文说明🚫 未接管：fork 托管要求 PR 作者在本仓库持有 write 及以上权限（作者 |
chiga0
left a comment
chiga0
left a comment
There was a problem hiding this comment.
Review — APPROVE
Reviewed at HEAD d18f38a. No Critical or Major issues. This PR has been through an unusually thorough review cycle, and I independently confirmed that every previously-raised correctness concern is resolved in the current HEAD, and found no new blockers.
What I verified independently
VP decision is correctly startup-scoped and coupled to Ink's alternateScreen lifetime. startInteractiveUI computes useVP = shouldUseVirtualViewport(...) once, passes it as initialUseVirtualViewport, and feeds Ink's render({ alternateScreen: useVP }) (line 236). AppContainer locks the same value via useState(() => initialUseVirtualViewport ?? shouldUseVirtualViewport(...)) — the ?? correctly respects an explicit false. This eliminates the TOCTOU where a live settings read could desync the React render path from the already-entered alternate screen. requiresRestart: true in the schema matches this semantics.
The safety gates are present and correct. shouldUseVirtualViewport = terminalInteractive && (useTerminalBuffer ?? true) && !screenReader, and isInteractiveTerminal = stdoutIsTTY && !isCiEnvironment && TERM !== 'dumb'. This is the fix for the piped/CI history-loss regression (qwen | tee and CI now stay on the legacy append-only <Static> path). Screen-reader mode is excluded from both the VP path and the alternate screen. CI detection (CI / CONTINUOUS_INTEGRATION / CI_* with active values) is a reasonable heuristic, and the non-TTY check provides a second layer of protection.
Consumers are consistently migrated. All mouse-interaction consumers (useMouseEvents, HistoryItemDisplay, InputPrompt, BaseSelectionList) now read useVirtualViewport(settingsFallback) — context-first with a settings fallback for out-of-provider/test rendering. uiState.useTerminalBuffer is the same startup-locked state variable (AppContainer line 1160 → uiState literal line 4268), so InputPrompt's liveAgentPanelMaxRows agrees with the context value.
Test coverage is solid. terminal-buffer.test.ts covers the shouldUseVirtualViewport truth table and isInteractiveTerminal (TTY/dumb/CI). The drift-guard assertion now reads getSettingsSchema()...useTerminalBuffer.default, so a schema/runtime default divergence fails the test rather than relying on the comment. The clearCiEnv test utility correctly saves/restores CI env vars so VP tests don't false-fail on GitHub Actions.
Cross-Validation
| Prior finding | Status | My check |
|---|---|---|
Screen-reader accessibility regression (dropped <Static>) |
Fixed (90ea976) | Confirmed — !screenReader gate in the shared helper |
TOCTOU: live useTerminalBuffer vs one-shot alternateScreen |
Fixed (8f5eb98) | Confirmed — startup-locked useState, requiresRestart: true |
| Piped/CI sessions lose history (ink emits only last frame) | Fixed (8632a62) | Confirmed — isInteractiveTerminal gate present in both call sites |
| Redundant/harmful alternate-screen exit handler (aborts later exit listeners) | Fixed (604e733, removed) | Confirmed — no PR-owned exit listener remains |
?? true dual source of truth with schema default |
Mitigated (8632a62) | Confirmed — test now pins against the schema default |
| Tests false-fail in CI (isTTY=true but CI env set) | Fixed (ci-env.ts + 975d216/81daecf) | Confirmed — clearCiEnv utility present |
debugLogger breadcrumb for VP decision |
Open (non-blocking) | Agree it's discretionary; not a correctness issue |
ThinkingViewer useAlternateScreen test |
Declined by author | Reasonable — value derives from the covered startup-locked state |
Minor observations (non-blocking)
InputPromptreadsuiState.useTerminalBufferforliveAgentPanelMaxRowsbutuseVirtualViewport(...)formouseInteractionsEnabled. Both resolve to the same value in-app; the mixed access pattern is only a readability nit.- The
useTerminalBuffersetting name remains counterintuitive (true= virtualized viewport, not host scrollback), but that naming is pre-existing and out of scope here.
hi, cloud you please solve the CI failure? |
qwen-code-dev-bot
left a comment
qwen-code-dev-bot
left a comment
There was a problem hiding this comment.
CI failure is EACCES infrastructure issue on ecs-qwen-runner-sg-12, not a code problem. Fix in PR #7931.
|
Released in v0.21.1. |
8 participants
What this PR does
Turns virtualized history on by default for interactive CLI sessions. New users and existing users without an explicit setting now get the in-app scrollable history viewport, while users who prefer host terminal scrollback can still opt out by setting
ui.useTerminalBuffertofalse.The settings schema, generated IDE schema, tests, and user/design docs are updated so the runtime behavior and documented behavior match.
Why it's needed
The legacy terminal scrollback rendering path can visibly flash or make the host terminal scrollbar jump during long sessions, refreshes, resize settle repaint, compact-mode toggles, and similar TUI updates. The virtualized history path already avoids the physical terminal clear/replay behavior that causes this class of flicker, but it was still opt-in, so most users continued to hit the old path unless they discovered the setting.
Defaulting to the virtualized path makes the stable behavior the common path, reduces repeated flicker reports, and keeps a small compatibility escape hatch for terminals or workflows that still prefer native scrollback.
Reviewer Test Plan
How to verify
Start the interactive CLI with no
ui.useTerminalBuffersetting and confirm it opens in the virtualized history flow. Scrolling history should use Shift+Up/Down, PgUp/PgDn, Ctrl+Home/End, or the mouse wheel instead of native terminal scrollback.Start the interactive CLI with
ui.useTerminalBuffer: falseand confirm the legacy host scrollback path is still available.Run the focused unit tests and type/build checks listed below.
Evidence (Before & After)
Before: with no explicit setting, the CLI defaulted to the legacy terminal scrollback path, so refreshes could clear/replay terminal history and produce visible scrollbar jumps in long sessions.
After: a local PTY smoke test launched the built CLI twice. With the setting unset, the captured ANSI stream entered the alternate screen and reached the input prompt. With
ui.useTerminalBuffer: false, the captured ANSI stream did not enter the alternate screen and still reached the input prompt.Verification commands run locally:
Tested on
Environment (optional)
Local source checkout, built CLI, sandbox disabled for the PTY smoke. The smoke did not send a model request; it only verified startup rendering and the terminal mode selected by default versus explicit opt-out.
Risk & Scope
ui.useTerminalBuffertofalseto keep the legacy host scrollback behavior.Linked Issues
N/A
中文说明
What this PR does
这个 PR 将交互式 CLI 的虚拟化历史视图默认开启。新用户和没有显式配置该选项的现有用户会默认使用应用内可滚动历史视口；如果用户更偏好宿主 terminal scrollback，仍然可以通过设置
ui.useTerminalBuffer为false选择旧路径。配置 schema、生成的 IDE schema、测试以及用户/设计文档都同步更新，保证运行时行为和文档描述一致。
Why it's needed
旧的 terminal scrollback 渲染路径在长会话、刷新、resize settle repaint、compact mode 切换等 TUI 更新场景下，可能出现明显闪屏或宿主 terminal 滚动条跳动。虚拟化历史路径已经避免了导致这类 flicker 的物理清屏和历史重放行为，但之前仍是 opt-in，大多数用户如果没有发现这个设置，依然会走旧路径。
把虚拟化路径设为默认，可以让更稳定的行为成为常规路径，减少重复的闪屏反馈，同时保留一个很小的兼容逃生口，给仍然偏好原生 scrollback 的终端或工作流使用。
Reviewer Test Plan
How to verify
在没有设置
ui.useTerminalBuffer的情况下启动交互式 CLI，确认它进入虚拟化历史流程。历史滚动应使用 Shift+Up/Down、PgUp/PgDn、Ctrl+Home/End 或鼠标滚轮，而不是依赖原生 terminal scrollback。在
ui.useTerminalBuffer: false的情况下启动交互式 CLI，确认旧的宿主 scrollback 路径仍然可用。运行上面列出的聚焦单测、typecheck 和 build 检查。
Evidence (Before & After)
Before：没有显式配置时，CLI 默认走旧的 terminal scrollback 路径，因此刷新时可能清屏并重放 terminal 历史，在长会话中表现为可见滚动条跳动。
After：本地 PTY smoke test 启动了两次构建后的 CLI。配置未设置时，捕获到的 ANSI 流进入了 alternate screen，并正常到达输入提示。显式设置
ui.useTerminalBuffer: false时，捕获到的 ANSI 流没有进入 alternate screen，也正常到达输入提示。本地执行的验证命令：
Tested on
Environment (optional)
本地源码 checkout，使用构建后的 CLI，PTY smoke 中关闭 sandbox。这个 smoke 没有发送模型请求，只验证启动渲染以及默认配置和显式 opt-out 分别选择的 terminal mode。
Risk & Scope
ui.useTerminalBuffer为false可以保留旧的宿主 scrollback 行为。Linked Issues
N/A