fix(clipboard): attribute multi-event OS echoes per write instead of one-shot - #51
Open
moekyo wants to merge 8 commits into
Open
fix(clipboard): attribute multi-event OS echoes per write instead of one-shot#51moekyo wants to merge 8 commits into
moekyo wants to merge 8 commits into
Conversation
…one-shot A single programmatic clipboard write can surface as several watcher events (OS re-encode, double change notification, clipboard-manager re-assertion). The one-shot next-change fallback absorbed only the first; the second was mis-attributed as LocalCapture and dispatched back to the sender, sustaining an A<->B image echo loop. The fallback now carries an echo budget per write (two credits for remote pushes, one for local restores), a content match spends one credit of the paired fallback instead of deleting it, fallback consumption is scoped to the same content kind (image/text/files/ rich-text), and the leftover credit expires ECHO_TAIL_TTL after the first echo so an unrelated later copy is not swallowed. Local restore semantics are unchanged (one-shot).
Contributor
|
Important
This repository does not receive automatic reviews because it has fewer than 10 stars. ⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Plus Run ID: Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Author
真机验证记录（macOS + Windows 双端）
|
Author
|
@mkdir700 麻烦 review 一下这个修复。问题：两台已配对设备只截一张图 → 本机面板同一张图不断新增、对端不断收到同一张图（捕获/同步风暴）。根因：SelfWriteLedger 一次性归因无法吸收一次写入产生的多个 OS 事件，第二个事件被判 LocalCapture 回传对端形成 A↔B 回环。已附单元测试、跨层探针和 macOS+Windows 真机验证记录。 |
Codecov Report✅ All modified and coverable lines are covered by tests. 📢 Thoughts on this report? Let us know! |
…g recovers A crash mid-admission (e.g. Windows blue screen during a join) leaves a durable 'admission in progress' record in the admission store. The last-resort /encryption/factory-reset path wiped key material and peer state but never cleared that record, so every later pairing attempt failed on both sides ( / with AdmissionInProgress) and the user was locked out of re-pairing. factory_reset now clears the admission store through the same clear_admission_state entry used by the device-management reset; a failed admission-store reset surfaces as AdmissionStateResetFailed instead of being swallowed.
Author
|
补充第二个修复（同分支）：factory reset（忘记口令兜底路径）现在会清理持久化的 admission 状态 —— 崩溃/蓝屏中断加入流程留下的 "admission in progress" 记录此前会导致重置后仍无法重新配对，现已修复。 |
A crash or interruption during pairing leaves a durable pending admission attempt that blocks every later pairing attempt. Add a lightweight recovery operation that clears only the non-terminal attempts - terminals, membership history, consumed invitations and the device trust revision of an intact space are preserved. - admission_attempt_store: clear_pending_admissions (attempts.retain non-terminal only) - port: AdmissionAttemptRepositoryPort::clear_pending_admissions - convergence: WorkspaceConvergence::clear_pending_admissions + ProfileWorkspaceConvergence::clear_stale_admission - engine: Operation::ClearStaleAdmission -> OperationResult::StaleAdmissionCleared
… swallowing Windows joiner fails at 'prepare Candidate: space transition storage failed' because init_db_pool (SqliteConnection::establish + WAL pragma) errors are collapsed to the unit AdmissionSpaceTransitionError::Storage, hiding the real diesel/libsqlite error. Log the underlying error plus the target database path and space id before collapsing; same for the manifest-store load in prepare_if_needed.
The 'space transition storage failed' on Windows joiner is collapsed at three unlogged points, hiding the real sqlite/diesel error: - prepare_if_needed now logs the branch decision (source space vs target space, manifest presence) - prepare_source failures now log the underlying String error plus generation/attempt ids - snapshot_to_bytes logs the diesel error for VACUUM INTO (e.g. 'cannot VACUUM from within a transaction', 'database is locked') with the scratch path and SQL
…torage
The Windows joiner failure ('space transition storage failed') was
invisible because every underlying sqlite/diesel/fs error was collapsed
into a unit Storage variant and the diagnostic tracing line could be
dropped by the lossy non-blocking writer. Make the error itself carry
the detail: Storage is now Storage(String) via a storage() constructor,
so the real failure text survives map_space_transition_error ->
WorkspaceConvergenceError::AdmissionStorage and appears in the
join-space error line and UI regardless of log writer behavior.
Windows joiner failed at 'space transition storage failed: open generation parent': std::fs::File::open on a directory requires FILE_FLAG_BACKUP_SEMANTICS, which the standard API does not pass, so opening the generation directory for fsync always fails with access denied on Windows (macOS/Linux allow read-only directory opens). Use the same pattern as active_space_manifest_store::sync_parent_directory: Unix opens+syncs the directory; Windows skips it (renames are already durable via MOVEFILE_WRITE_THROUGH).
Author
|
Windows joiner 修复验证通过（7efa66ec） 问题：Windows 作为 joiner 加入空间必败 根因： 修复：复用 验证：Windows 实机以 joiner 身份用邀请码加入 → 成功打通（此前 5 个版本必败）。 |
Member
|
感谢你的 PR，这两天我在重新梳理 application 层（AI 写的太乱了），不可避免会有冲突，所以 PR 暂时无法合并 |
Author
我自己先打包用着了，不急的 :) |
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
2 participants
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.
问题类型
结构性缺陷（origin 归因语义：一次性消费 → 按单次写入的回显集）。
根因
一次程序化剪贴板写入可能产生多个 watcher 事件（OS re-encode、双变更通知、剪贴板管理器重断言）。原
InMemorySelfWriteLedger的 next-change fallback 只吸收第一个事件，第二个事件无 guard 命中 → 被判为LocalCapture→dispatch_local_capture回传原设备 → 对端 apply 后再写 OS 剪贴板 → 形成 A↔B 图片回环。跨层身份不一致：watcher 用解码像素指纹去重，ledger 用原始字节 hash，re-encode 后 ByContent 永不命中，完全依赖这个一次性 fallback。用户现象：两台已配对设备，只截一张图 → 本机面板同一张图不断新增、对端不断收到同一张图（捕获/同步风暴）。
方案
image:/text:/files:/rich-text:前缀）匹配，无关类别的事件不再抢占。ECHO_TAIL_TTL（5s），后续无关复制不被吞。验证
cargo test -p uc-infra --lib clipboard::change_origin：12 通过（含新增回环回归用例）。cargo check --workspace --all-targets --locked（585 crates）、cargo fmt --all -- --check、node scripts/architecture/check-engine-repository.mjs通过。uc-infra全量测试中rendezvous::client::tests::transport_failure_maps_transport_variant在本机失败，已在基础提交（31c149c）上复现，与本改动无关（环境相关）。文档
已按仓库规范在
docs/architecture/architecture-bible.md文档维护记录中登记本次语义变化。