Repository navigation
refactor(#2607): remove compress.stripImages — default media folding replaces it - #2688
Merged
Merged
Conversation
…replaces it Default archivable-media folding (#2640) made the opt-in wire-side image strip redundant, and the strip itself broke the prompt cache every turn on three of four wires (sliding cutoff moves the byte boundary per turn; only anthropic had the fold-anchored boundary). Media handling now has exactly one path: fold + archive + decompress({ imageRef }) recovery. - kernel: strip-images.ts (stripHistoricalImages) retired; wire barrel and wire-google tests updated; acp-kernel 0.0.113 -> 0.0.114 (kernel-change: removing a public wire export that exists only for the removed host policy; no compression behavior changes) - host: strip wiring, fold-anchored cutoff, and DEFAULT_STRIP_IMAGES_KEEP_RECENT removed from handle.ts/server.ts/compress-settings.ts; image-restore.ts keeps the always-on index + spill + decompress machinery (that is the default-fold recovery path, not the strip's) - config: compress.stripImages / compress.stripImagesKeepRecent dropped from the type, whitelist, and parser; old configs get a one-time INERT warning (warnInertStripImages) instead of silent no-op - docs: seeds + generated tables regenerated; EN/ZH prose sections replaced with removal notes; decompress error texts no longer reference the flag - tests: strip-images / strip-images-compact / strip-log-observability / strip-cutoff-anchor suites deleted; compress-settings cascade/parse cases removed; new strip-images-removal.test.ts pins the INERT notice + silence Closes #2608 (archival-failure-still-strips existed only on the strip path; with the strip gone there is nothing to fail)
Contributor
📦 Built Package ArtifactBranch: Option A — Install this exact build (recommended)npm install -g billion-context@0.0.2903-pr.2688.46853b8This exact version is immutable and reproducible. The rolling Option B — Download artifact
tar xzf billion-context-pr2688.tgz
npm install -g packageInstall instructions are refreshed on each push. |
Review nit from the PR review: the #618 note above prepareResponsesCompact still described the removed conditional re-serialization ("forward the re-serialized post-strip work instead so dropped images don't ride along"). With the strip gone the call always forwards bodyBuffer unchanged, so the note now matches the code (byte-identical passthrough when no bili compaction item is present).
Owner
Author
[bot] 🏷 Review 完成：可合（等新 head 的 CI 全绿）。我拉了分支逐文件通读 diff、本地定向验证、并对最新 master 做了 rebase 检查。结论：方向正确、执行干净，未发现回归；顺手修了一个小问题（stale 注释）已推到 PR 分支。 验证证据
顺手小修（独立标注 commit
既有行为回归清点（逐项）
高危域点名核查：自动更新/自更新路径未触碰；版本与 lockfile — 仅 结构化八问
补充：kernel 改动属 human-only review 层（碰了 判定：可合 —— 待新 head |
Owner
Author
[bot] 🏷 CI 全绿确认：rebase 后的新 head 之前 最终判定维持：可合。合并请由 @ranxianglei 本人执行（kernel 改动属 human-only 层）。合并后 #2608 随之关闭。 |
Contributor
📦 Merged Build PublishedThis merge to master was published as npm install -g billion-context@0.0.236-master.9d1b9d2 # this exact build
npm install -g billion-context@master # always the newest merged build
|
This was referenced Oct 10, 2026
Merged
ranxianglei
pushed a commit
that referenced
this pull request
Oct 10, 2026
…semantics Supplement to 990cbe8 (which fixed only the tsc errors): under post-#2688 semantics there is no wire-side image reduction left, so the two tests that asserted the removed stripImages recovery path (200 + reduced rebuilt body) fail at runtime with the forward-stage 413. Adapt them to assert the actual contract: over-budget rebuilt bodies stop at stage=forward / request_too_large without touching upstream, whether images sit in messages or tool outputs; the admission-slot flow ends at the same forward guard once a slot is released. Also fix reference/request-body-budget.md (en+zh), which still told users to configure the removed compress.stripImages. Verified locally: tsc --noEmit clean; node --import tsx --test --test-concurrency=1 tests/image-request-budget.test.ts tests/request-body-budget.test.ts -> 10/10 pass.
This was referenced Oct 10, 2026
YunzhouLi-hub
pushed a commit
to YunzhouLi-hub/billion-context
that referenced
this pull request
Oct 11, 2026
…-request-budget test ranxianglei#2688 removed compress.stripImages/stripImagesKeepRecent from CompressSettings, but ranxianglei#2613's image-request-budget test (merged after ranxianglei#2688) still referenced them, breaking master typecheck. Drop the dead keys from the fixture config and the inert stripImages=false toggle line; keep every assertion intact (over-budget rebuilt request -> forward-stage 413, no upstream contact). Zero behavioral change.
ranxianglei
added a commit
that referenced
this pull request
Oct 11, 2026
…rebase (#2671) Rebase onto master (47 commits) surfaced two machine-dependent seams and one real timing bug: 1. Test hermeticity (storage-corpus / storage-unified / golden-storage-drive / web-sessions): these suites point SessionStore at the REPO corpus dirs or tmp dirs and assert on directory contents. On a developer machine whose real billion-context.json sets storage.unified=full (exactly the dogfood setup this PR exists to enable), the config-file tier leaked into every test that cleared the env expecting legacy: stray index.db files inside tests/golden/storage-corpus (breaking the no-orphans pin and being cpSync'd into downstream tmp dirs), sidecars inside web-sessions tmp dirs tripping the #1937/#2180 foreign-content alarms. Pin BILI_CONFIG_FILE to a missing tmp path and clear BILI_STORAGE_UNIFIED at module top of each suite — same pattern master's #2673 established for XDG state. 2. Eager storage-mode stamp: the kernel write chain may invoke the record builder on a later tick than the caller (serialized write chains, #2674). guardedBuild captured effectiveFileStorageMode() inside the deferred thunk, so a save enqueued under one mode could be STAMPED under whatever the env said when a background chain reached the build. Capture the stamp eagerly at enqueue time and thread it into buildRecord. 3. storage-render goldens regenerated (stateDigest + metadata section only; state/blockContents/stats sections unchanged): master #2688 removed compress.stripImages from the effective config, which is mirrored into persisted session metadata.effectiveConfig. Battery: 69/69 (golden-storage-render, storage-corpus, storage-unified, web-sessions, config-file-cache, config-surface-guard, admin-origin, image-request-budget), tsc clean, unused-export gate clean (2 waived). storage-unified passes even with a hostile BILI_STORAGE_UNIFIED=full exported into the test process.
ranxianglei
added a commit
that referenced
this pull request
Oct 11, 2026
…rebase (#2671) Rebase onto master (47 commits) surfaced two machine-dependent seams and one real timing bug: 1. Test hermeticity (storage-corpus / storage-unified / golden-storage-drive / web-sessions): these suites point SessionStore at the REPO corpus dirs or tmp dirs and assert on directory contents. On a developer machine whose real billion-context.json sets storage.unified=full (exactly the dogfood setup this PR exists to enable), the config-file tier leaked into every test that cleared the env expecting legacy: stray index.db files inside tests/golden/storage-corpus (breaking the no-orphans pin and being cpSync'd into downstream tmp dirs), sidecars inside web-sessions tmp dirs tripping the #1937/#2180 foreign-content alarms. Pin BILI_CONFIG_FILE to a missing tmp path and clear BILI_STORAGE_UNIFIED at module top of each suite — same pattern master's #2673 established for XDG state. 2. Eager storage-mode stamp: the kernel write chain may invoke the record builder on a later tick than the caller (serialized write chains, #2674). guardedBuild captured effectiveFileStorageMode() inside the deferred thunk, so a save enqueued under one mode could be STAMPED under whatever the env said when a background chain reached the build. Capture the stamp eagerly at enqueue time and thread it into buildRecord. 3. storage-render goldens regenerated (stateDigest + metadata section only; state/blockContents/stats sections unchanged): master #2688 removed compress.stripImages from the effective config, which is mirrored into persisted session metadata.effectiveConfig. Battery: 69/69 (golden-storage-render, storage-corpus, storage-unified, web-sessions, config-file-cache, config-surface-guard, admin-origin, image-request-budget), tsc clean, unused-export gate clean (2 waived). storage-unified passes even with a hostile BILI_STORAGE_UNIFIED=full exported into the test process.
Merged
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
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.
Removes
compress.stripImages/compress.stripImagesKeepRecententirely. Default media folding (#2640) made the opt-in wire-side image strip redundant, and the strip itself was a turn-over-turn prompt-cache breaker on three of four wires: its sliding cutoff (len - keepRecent) moves the outgoing byte boundary every turn, so the cache forks at the most expensive content each round (only anthropic had the fold-anchored boundary). Media handling now has exactly one path: fold + archive +decompress({ imageRef })recovery.kernel-change: acp-kernel 0.0.113 → 0.0.114 —
stripHistoricalImages(kernel/src/wire/strip-images.ts) was a public wire export that existed only for this removed host policy; no compression behavior changes.Why now (evidence, not taste)
compress.stripImages默认开启——否则带图消息不可压缩会形成只升不降的输入地板撑爆上下文 #2607's follow-ups landed: archivable media folds by default, pixels spill to<state>/retrieve/img/<session>/BEFORE the fold,decompress({ imageRef })restores them (Archive-then-fold: default folding for archivable media (Closes #2607, Fixes #2609) #2640, kernel 0.0.113).Changes
strip-images.tsdeleted; wire barrel export removed;wire-googlestrip tests removed; version bump.foldAnchoredCutoff+DEFAULT_STRIP_IMAGES_KEEP_RECENTremoved (handle.ts,server.ts,compress-settings.ts).image-restore.tskeeps the always-on index/spill/decompress machinery — that is the default-fold recovery path.[acp-config] … is INERT … was removedwarning (warnInertStripImages, pattern ofwarnInertResignPassthrough).strip-images/strip-images-compact/strip-log-observability/strip-cutoff-anchorsuites deleted; compress-settings cascade/parse cases removed; newstrip-images-removal.test.tspins the INERT notice (fires once, names the keys) and that clean configs stay silent.Verification
check:unused-exportsclean;gen-config-docs check-coverageokCloses #2608