Skip to content

perf(chat): bound the background warm with the same window walk - #16005

Merged
bolichen97 merged 1 commit into
mainfrom
perf/bounded-warm-walk
Oct 2, 2026
Merged

bolichen97 merged 1 commit into
mainfrom
perf/bounded-warm-walk

Conversation

@buluoray

@buluoray buluoray commented Oct 1, 2026 •

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: a background session's cache warm (warmSlotCache) stops re-downloading the whole transcript on every background turn end and every reconnect once it holds more than 500 rows.

warmSlotCache runs on every off-screen chat_done (website/src/hooks/websocket/turnCompletion.ts) and on every reconnect. It still read the whole transcript in two cases: when the pane held more than one 500-row page (the count-matched limit declines past the handler's clamp), and when the bounded window missed a cached row. A live capture shows the result: SWITCH short=7546 lim=500 cached=8104 total=8046 -- an active session's background cache already held its full history, re-read each turn.

The merged #13236 replaced the same unbounded fallback in refreshSlot and switchSlot with a bounded backward walk (walkWindowBackTo), and listed the warm as out of scope. This PR builds on it. Billy Gerhard (@billygerhard) found and measured this same warm problem in #15029; this PR redoes only the warm fix on top of #13236's walk mechanism. Credit for finding and measuring it is his (co-author on the commit).

Why it matters

Each unbounded read makes the handler load, redact and serialize the whole session on one worker thread. A long session in a background pane pays that on every turn end, and a reconnect pays it for every pane on screen at once, which stalls other requests and WebSocket frames behind it.

Not a goal

What changed (motivation → approach → change)

Warm case Before After
Cache past 500 rows 1 unbounded read 1 page of 500, walk only if it misses
Window misses cached rows 1 bounded + 1 unbounded read walk older, at most 1 + 8 + 1 bounded reads
Cached row newer than the window's start (rewind, rewrite, stale response) unbounded retry 1 fresh read of the same bounded page
No identified row / unplaceable row unbounded unbounded (unchanged)

The count-matched limit returned undefined for two different reasons: the cache is too wide for the clamp, or its rows cannot be identified. Only the second needs the whole transcript. warmSlotCache in website/src/store/chat/slotRefresh.ts now runs the same rule with no ceiling to tell them apart, asks for one 500-row page when the cache is merely wide, and closes a coverage miss with walkWindowBackTo, exactly as refreshSlot and switchSlot call it. The reducer keeps the cached head above the walk's anchor, so no scrollback is lost.

A miss a walk older cannot close is one where the window anchors the cache but lacks a cached row after that anchor. The old unbounded retry doubled as a fresh read for that case, so it gets one fresh read of the same bounded page, and the reducer's existing shrink and same-count-rewrite checks decide what was removed. A walk that spends its 8-page cap without anchoring hands back a window disjoint from the cache; the warm marks it, and the reducer replaces the cache with that window instead of splicing the two with a silent hole between them. That replacement applies only to the exact cache the warm read: if anything wrote to the pane while the warm was in flight (a message sent into the background pane, a streaming chunk landing on an existing row, a newer warm or refresh), the cache's array identity changed, and the warm leaves the pane's rows and paging markers alone for the next warm to reconcile. A walk that stops on a window containing the whole cache is not marked, so rows newer than the page (a just-sent message, a client streaming copy) are still kept. Because every request is now bounded, the retained total is always the collapsed count; the raw-count bookkeeping the unbounded retry needed (comparableTotal) is gone from this path. website/src/store/chat/windowWalk.ts only gains a doc line naming the warm as a caller.

Backwards compatibility

Compatible: no API or stored shape changes; the warm sends the same request with a smaller limit. One behavior narrows: when more than 4,000 rows arrived since the last warm (past the walk's 8-page cap), the pane's cache becomes the newest 4,500 rows instead of the whole history; the older rows are one page-back away, the same cost #13236 documents for refreshSlot and switchSlot.

Tests

  • website/src/store/chatSlice.windowWalk.test.ts, new warmSlotCache block, on a multi-page fake handler:
    • an 8,000-row cache warms with exactly one 500-row request and keeps all 8,000 held rows plus the new ones;
    • a coverage gap walks 2 older pages plus the newest re-read, keeps cached rows 0..499 above the anchor, and never asks for row 0's page;
    • a walk that stops on a window containing the whole cache keeps a just-sent row at the tail;
    • a gap past the cap sends exactly 1 + 8 + 1 requests, none unbounded, and replaces the cache with the walked window (no spliced hole) while slotPaneHasMore stays true.
  • website/src/test/chatSlice.warmSlotCacheBound.test.ts: the rewind and same-count-rewrite cases now pin one fresh bounded read ([50, 50]) instead of an unbounded one; the raw-total retry tests are reworded for bounded totals, and mock queues that no longer get consumed are removed.
  • website/src/store/chatSlice.boundedRefetchShrink.test.ts: the oldest-111 cache now pins [111, 500, 500] instead of [111, undefined].
  • Reverting slotRefresh.ts reddens 9 of these tests, and dropping the reducer's replace-on-unanchored-walk branch reddens the cap test, including all three new ones. Locally: 192 related store/hook test files (4,201 tests) pass, tsc -p tsconfig.app.json and eslint clean.

Manual verification

N/A — unit coverage sufficient: the change is the limit and request sequence the thunk sends, which the tests assert directly against a fake handler with real paging.

Screenshots / video

Why no screenshot: store-only change to how many rows a background fetch requests; nothing rendered differs.

Related Issues

no linked issue: builds on the merged #13236; same problem as #15029, which is credited above.

Checklist

  • At most two commits (one is the norm), with a Conventional Commits title (feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)
  • Existing tests pass and new tests added for new functionality
  • Self-review completed; code follows project style guidelines
  • Documentation updated (if applicable)
  • No secrets, credentials, or internal references in the diff

@buluoray
buluoray requested a review from a team October 1, 2026 21:02
@buluoray
buluoray requested a review from a team as a code owner October 1, 2026 21:02
@buluoray
buluoray requested a review from chenmingwei23 October 1, 2026 21:02
@github-actions github-actions Bot added the readiness: checking Automated validation is still running label Oct 1, 2026
@github-actions

github-actions Bot commented Oct 1, 2026 •

Copy link
Copy Markdown
Contributor

UX Review (Fable 5) — ✅ PASS

UX-level review of bc47ebaa11a7e3f63fa008f4812350c7a3f42a7f — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

UX-Verdict: PASS

Verify a background pane past the walk cap still restores older scrollback via the normal page-back affordance after cache replacement.

[UX-REVIEWED] bc47eba

Description read

[DESCRIPTION-READ] dfe65152c8db8c968b76e501aeba4730996312aacc530866059ddf8520601475

That sha256 names everything this verdict read: the PR description, plus the 3 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions

github-actions Bot commented Oct 1, 2026 •

Copy link
Copy Markdown
Contributor

Design Review (Fable 5) — ✅ PASS

Design-level review of bc47ebaa11a7e3f63fa008f4812350c7a3f42a7f — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

Design-Verdict: PASS

Confirm the one UX narrowing: a >4,000-row gap now replaces a background pane's held scrollback with the walked window (page-back recovers it), matching #13236's documented cost.

Suggestions

  • The cacheAtDispatch array-identity guard (isDraft/original in the reducer) is the subtlest new coupling — if a future slotMessages writer ever lands outside Immer's draft path, the replace fires on a changed cache; a one-line invariant comment at the pane's write helpers would keep that contract visible to the next writer.

[DESIGN-REVIEWED] bc47eba

Description read

[DESCRIPTION-READ] a48a9e4268df5c07b9a91da65e0ed6fe0c2297fedb643407287dc5003d5d6b3e

That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions

github-actions Bot commented Oct 1, 2026 •

Copy link
Copy Markdown
Contributor

First Principles Review (Fable 5) — ✅ PASS

Premise-level review of bc47ebaa11a7e3f63fa008f4812350c7a3f42a7f — why this exists and whether the shipped surface is the smallest honest version. Updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

First-Principles-Verdict: PASS

Verify the one shipped narrowing: a >4,000-row gap now replaces the pane's held history with the newest ~4,500 rows — declared, and the same page-back cost #13236 recorded for the sibling readers.

Not justified as shipped

  • Item 7 — undeclared, rides along: the test store's middleware: ... serializableCheck: false, immutableCheck: false line is deleted ("Match the production store") in chatSlice.windowWalk.test.ts; test-only, no premise risk.

What this change ships

Inventory (8 items) — 7 justified

Intent: stop a background pane's cache warm from re-downloading the whole transcript once it holds more than one page — a FIX, with provenance (measured capture in the description, #15029, and 9 added tests that redden on revert).

  1. A pane past 500 rows warms with one bounded page, not the whole transcript — justified
  2. A coverage miss walks older bounded (≤1+8+1 requests) via existing walkWindowBackTo — justified
  3. Rewind/rewrite confirmation is one fresh bounded page instead of an unbounded retry — justified
  4. A gap past the walk's cap replaces the cache with the walked window instead of splicing a silent hole — justified
  5. That replacement yields to any concurrent write to the pane (cache identity guard) — justified
  6. comparableTotal raw-count bookkeeping deleted from the warm path (its unbounded premise is gone) — justified
  7. windowWalk tests now run the production store's default dev checks — undeclared, rides along (test-only)
  8. windowWalk.ts doc line names the warm as a third caller — justified

Consumer counts: walkUnanchored and cacheAtDispatch each have exactly 1 real consumer (the warmSlotCache.fulfilled reducer), singular form, not generalized. Unfixed siblings: grep fetchSlotDetail( shows the only remaining unbounded reads are the identity-guard cases the Goal's ## Not a goal carves out; refreshSlot/switchSlot were fixed in merged #13236. The rewritten test pins ([111, undefined], raw-total retries) pinned the unbounded shape this PR removes; the description declares each reversal and the premise's removal.

[FIRST-PRINCIPLES-REVIEWED] bc47eba

Description read

[DESCRIPTION-READ] 7893fe1bcec52747fe2c7a77f0872e890cad6ec0739555dc1adcd2e7f3c359ce

That sha256 names the PR description this verdict read. Recompute it with .github/scripts/pr-description-capture.sh; if it differs, the description changed after this verdict was formed and any finding drawn from the description is unproven.

@github-actions

github-actions Bot commented Oct 1, 2026 •

Copy link
Copy Markdown
Contributor

GPT 5.6 Review — ✅ no blocking findings

GPT 5.6 completed its review of bc47ebaa11a7e3f63fa008f4812350c7a3f42a7f and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] bc47eba

False positive or not applicable? A repository writer can comment:
/ai-review override gpt bc47ebaa11a7e3f63fa008f4812350c7a3f42a7f: <one-sentence reason>

@github-actions

github-actions Bot commented Oct 1, 2026 •

Copy link
Copy Markdown
Contributor

Opus 5 Review — ✅ no blocking findings

Reviewed bc47ebaa11a7e3f63fa008f4812350c7a3f42a7f — this comment is updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] bc47eba

Verdict parsed from the review's SHA-scoped output markers for commit bc47ebaa11a7e3f63fa008f4812350c7a3f42a7f.

False positive or not applicable? A repository writer can comment:
/ai-review override fable bc47ebaa11a7e3f63fa008f4812350c7a3f42a7f: <one-sentence reason>

@buluoray
buluoray force-pushed the perf/bounded-warm-walk branch from d195500 to 5c519a0 Compare October 1, 2026 22:44
@buluoray

buluoray commented Oct 1, 2026 •

Copy link
Copy Markdown
Contributor Author
  • span=8d3da0b34990: an unanchored walk discards rows appended while the warm is in flight

Outcome: fixed in 5c519a0.
The thunk now records the cache length it read at dispatch (cacheLengthAtDispatch) whenever it marks a walk unanchored. The reducer still replaces the cache with the walked window, then appends the rows past that boundary that the page lacks (tailNotInPage). So a message sent into the background pane, or a new streaming reply, survives the replacement. Nothing is rescued when the cache got shorter than the boundary or the server count shrank.
Regression test: rescues rows appended during an unanchored warm of %i cached rows in website/src/store/chatSlice.windowWalk.test.ts, at 50 and 5000 held rows. It checks 1 + 8 + 1 bounded requests, the appended user and streaming rows kept after the window, an echo already on the page not duplicated, and no rescue for a shortened cache or a server shrink. Both cases fail when the rescue is disabled.

@buluoray

buluoray commented Oct 1, 2026 •

Copy link
Copy Markdown
Contributor Author
  • span=58d5ab273dcf: zero consumers, dead comparableTotal reader in the warm reducer

Outcome: fixed in 5c519a0, by deleting the dead reader.
No warm payload sets comparableTotal (its only producer is switchSlot in slotSwitch.ts), so the warm reducer now compares and retains the payload's own total, and the comment that described the removed unbounded retry is gone. The retainServerTotal bounded-read flag is unchanged in effect: comparable !== undefined was always false on this path.

@buluoray

buluoray commented Oct 1, 2026 •

Copy link
Copy Markdown
Contributor Author
  • span=5c649519b069: coversCache / cutIdx < 0 re-derive the walk's own stop predicate

Outcome: kept as is in this PR, with a follow-up proposed.
The duplication is real. Removing it needs walkWindowBackTo to return its verdict (anchored, superset, or cap spent), but the frozen goal lists changing walkWindowBackTo under "Not a goal" (it shipped in #13236 and has two other callers). The twin uses the same anchor set and the same idAnchorsOneRow test as the walk, and the cap-exhaustion test pins the case where the two must agree. Follow-up: give the walk an anchored outcome and have the warm read it. A maintainer can lift the exclusion if they want that done here.

@buluoray

buluoray commented Oct 1, 2026 •

Copy link
Copy Markdown
Contributor Author
  • span=8a338190843f: have the walk return an anchored outcome flag (follow-up)

Outcome: accepted as the follow-up, not made in this PR.
This is the same change the First Principles item asks for. walkWindowBackTo is outside this PR's goal, so the warm keeps its matched re-derivation for now. The follow-up is to return the walk's anchor verdict and delete the three hand-synchronized copies.

@buluoray

buluoray commented Oct 1, 2026

Copy link
Copy Markdown
Contributor Author
  • span=8d3da0b34990: a stale unanchored warm can overwrite a newer transcript

Outcome: fixed in 1d07cfe.
The reducer's existing out-of-order test (staleTotal: this payload's warmSeq is older than the one that set the retained total) now also gates the unanchored replacement. When an older cap-exhausting warm settles after a newer one, it does not call writeSlotPage, so the newer warm's rows, slotPaneHasMore and slotPaneBounded stay as written. retainServerTotal already ignores the older sequence, and the run-state write was already ordered by warmSeq.
Regression test: keeps the newer cache and paging markers when an older unanchored warm settles last in website/src/store/chatSlice.windowWalk.test.ts. It fails when the guard is removed: the stale 4,050-row window replaces the newer cache.

warmSlotCache read a background pane's whole transcript whenever the pane
held more than one page (the count-matched limit declines past the
handler's 500-row clamp) and whenever the bounded window missed a cached
row. It runs on every off-screen chat_done and on every reconnect, so a
long session re-downloaded its full history on each background turn end.

A cache past the clamp now asks for one clamp-sized page, and a coverage
miss is closed with walkWindowBackTo, as refreshSlot and switchSlot do.
A miss the walk cannot close (a cached row newer than the window's
start: a rewind, a rewrite, or a stale response) is settled with one
fresh read of the same bounded page. A walk that spends its page cap
without anchoring replaces the cache instead of being spliced onto it,
so no hole is published. Every request on the path is
bounded, so the retained total is always the collapsed count. The
identity guards (no identified row, an unplaceable row) keep the
unbounded shape.

Co-authored-by: Billy Gerhard <billygerhard@gmail.com>
@buluoray
buluoray force-pushed the perf/bounded-warm-walk branch from 1d07cfe to bc47eba Compare October 2, 2026 00:40
@buluoray

buluoray commented Oct 2, 2026

Copy link
Copy Markdown
Contributor Author
  • span=8d3da0b34990: unanchored warm drops chunks added to an existing streaming row

Outcome: fixed in bc47eba by restructuring, not by a third special case.
Rounds 1-3 on this span were one defect: the unanchored branch replaced the cache from a dispatch-time snapshot, and any write to the pane during the flight invalidated that snapshot. The round-1 length-boundary rescue and the round-2 staleTotal skip are removed. A single rule replaces them: the thunk passes the cache array it read (cacheAtDispatch), and the reducer writes an unanchored page only when the slot's cache is still that same array (original() under Immer). Immer gives the array a new identity on an append, an in-place row edit such as a streaming chunk, and a newer warm's page, so all three now leave the pane's rows and paging markers alone for the next warm to reconcile.
Tests in website/src/store/chatSlice.windowWalk.test.ts: user row appended mid-warm (50 and 5,000 held rows), an in-place chunk through the real sseChatMessage reducer, and an older warm settling after a newer one. All four fail when the identity check is disabled. The untouched-cache case still replaces after 1 + 8 + 1 bounded requests with no hole.

@buluoray

buluoray commented Oct 2, 2026

Copy link
Copy Markdown
Contributor Author
  • span=a5c5e7618e43: invariant comment for the cacheAtDispatch identity guard

Outcome: acknowledged and not changed in this PR. The verdict on this head is PASS, and this is an advisory suggestion.
The guard fails closed. If a future slotMessages writer bypassed Immer and mutated the array in place, the identity would still match and the unanchored warm would replace the cache, which is the behavior before this round, not data loss beyond it. Every current writer goes through Immer reducers. The comment at the guard already states the rule ("may replace only the cache it read; Immer changes the array identity for appends, nested row edits, and replacement pages"). Adding a note at each write helper can go in a follow-up if a non-Immer writer ever appears.

@buluoray

buluoray commented Oct 2, 2026

Copy link
Copy Markdown
Contributor Author
  • span=0d24b3ecd7be: test store's serializableCheck/immutableCheck opt-out removed

Outcome: kept on purpose. Test-only, and it has a reason.
The production store (website/src/store/index.ts) calls configureStore with the default middleware, so the dev-mode serializable and immutable checks run there. The new warm payload carries cacheAtDispatch, a reference to the cached array, and the reducer compares identity under Immer. Running the window-walk suite with the same default checks proves that payload passes both checks and is not mutated, which the old opt-out would have hidden. All 16 tests in the file pass with the defaults.

@github-actions github-actions Bot added readiness: passed Eligible automated validation passed for the current revision and removed readiness: checking Automated validation is still running labels Oct 2, 2026

@bolichen97 bolichen97 left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Approved: PR Readiness green on this head, merges cleanly into current main, no outstanding blocking review.

@bolichen97
bolichen97 merged commit f93b53a into main Oct 2, 2026
97 checks passed
@bolichen97
bolichen97 deleted the perf/bounded-warm-walk branch October 2, 2026 05:05
@github-actions github-actions Bot removed the readiness: passed Eligible automated validation passed for the current revision label Oct 2, 2026
billygerhard added a commit to billygerhard/KiroCrew that referenced this pull request Oct 2, 2026
Rebased onto kirodotdev#13236 and kirodotdev#16005, which bound refreshSlot, switchSlot and
warmSlotCache with a clamp plus an older-walking window. This keeps that walk
on all three and adds only what a kept head needs on top of it: a marked view
or cache (a serving change) reads whole instead of walking, since the walk
keeps rows above its anchor verbatim, a walked or overlapping head still
offering a sign-in link reads whole, and a background warm that anchors
nothing in a cache with a raw retained count reads whole once so a remote
clear is still read as a shrink. The refreshSlotBound over-ceiling test is
kirodotdev#13236's own.
billygerhard added a commit to billygerhard/KiroCrew that referenced this pull request Oct 2, 2026
Rebased onto kirodotdev#13236 and kirodotdev#16005, which bound refreshSlot, switchSlot and
warmSlotCache with a clamp plus an older-walking window. This keeps that walk
on all three and adds only what a kept head needs on top of it: a marked view
or cache (a serving change) reads whole instead of walking, since the walk
keeps rows above its anchor verbatim, a walked or overlapping head still
offering a sign-in link reads whole, and a background warm that anchors
nothing in a cache with a raw retained count reads whole once so a remote
clear is still read as a shrink. The refreshSlotBound over-ceiling test is
kirodotdev#13236's own.
billygerhard added a commit to billygerhard/KiroCrew that referenced this pull request Oct 2, 2026
Rebased onto kirodotdev#13236 and kirodotdev#16005, which bound refreshSlot, switchSlot and
warmSlotCache with a clamp plus an older-walking window. This keeps that walk
on all three and adds only what a kept head needs on top of it: a marked view
or cache (a serving change) reads whole instead of walking, since the walk
keeps rows above its anchor verbatim, a walked or overlapping head still
offering a sign-in link reads whole, and a background warm that anchors
nothing in a cache with a raw retained count reads whole once so a remote
clear is still read as a shrink. The refreshSlotBound over-ceiling test is
kirodotdev#13236's own.
billygerhard added a commit to billygerhard/KiroCrew that referenced this pull request Oct 2, 2026
Rebased onto kirodotdev#13236 and kirodotdev#16005, which bound refreshSlot, switchSlot and
warmSlotCache with a clamp plus an older-walking window. This keeps that walk
on all three and adds only what a kept head needs on top of it: a marked view
or cache (a serving change) reads whole instead of walking, since the walk
keeps rows above its anchor verbatim, a walked or overlapping head still
offering a sign-in link reads whole, and a background warm that anchors
nothing in a cache with a raw retained count reads whole once so a remote
clear is still read as a shrink. The refreshSlotBound over-ceiling test is
billygerhard added a commit to billygerhard/KiroCrew that referenced this pull request Oct 2, 2026
Rebased onto kirodotdev#13236 and kirodotdev#16005, which bound refreshSlot, switchSlot and
warmSlotCache with a clamp plus an older-walking window. This keeps that walk
on all three and adds only what a kept head needs on top of it: a marked view
or cache (a serving change) reads whole instead of walking, since the walk
keeps rows above its anchor verbatim, a walked or overlapping head still
offering a sign-in link reads whole, and a background warm that anchors
nothing in a cache with a raw retained count reads whole once so a remote
clear is still read as a shrink. The refreshSlotBound over-ceiling test is
billygerhard added a commit to billygerhard/KiroCrew that referenced this pull request Oct 2, 2026
Rebased onto kirodotdev#13236 and kirodotdev#16005, which bound refreshSlot, switchSlot and
warmSlotCache with a clamp plus an older-walking window. This keeps that walk
on all three and adds only what a kept head needs on top of it: a marked view
or cache (a serving change) reads whole instead of walking, since the walk
keeps rows above its anchor verbatim, a walked or overlapping head still
offering a sign-in link reads whole, and a background warm that anchors
nothing in a cache with a raw retained count reads whole once so a remote
clear is still read as a shrink. The refreshSlotBound over-ceiling test is
billygerhard added a commit to billygerhard/KiroCrew that referenced this pull request Oct 2, 2026
Rebased onto kirodotdev#13236 and kirodotdev#16005, which bound refreshSlot, switchSlot and
warmSlotCache with a clamp plus an older-walking window. This keeps that walk
on all three and adds only what a kept head needs on top of it: a marked view
or cache (a serving change) reads whole instead of walking, since the walk
keeps rows above its anchor verbatim, a walked or overlapping head still
offering a sign-in link reads whole, and a background warm that anchors
nothing in a cache with a raw retained count reads whole once so a remote
clear is still read as a shrink. The refreshSlotBound over-ceiling test is
billygerhard added a commit to billygerhard/KiroCrew that referenced this pull request Oct 2, 2026
Rebased onto kirodotdev#13236 and kirodotdev#16005, which bound refreshSlot, switchSlot and
warmSlotCache with a clamp plus an older-walking window. This keeps that walk
on all three and adds only what a kept head needs on top of it: a marked view
or cache (a serving change) reads whole instead of walking, since the walk
keeps rows above its anchor verbatim, a walked or overlapping head still
offering a sign-in link reads whole, and a background warm that anchors
nothing in a cache with a raw retained count reads whole once so a remote
clear is still read as a shrink. The refreshSlotBound over-ceiling test is
billygerhard added a commit to billygerhard/KiroCrew that referenced this pull request Oct 3, 2026
Rebased onto kirodotdev#13236 and kirodotdev#16005, which bound refreshSlot, switchSlot and
warmSlotCache with a clamp plus an older-walking window. This keeps that walk
on all three and adds only what a kept head needs on top of it: a marked view
or cache (a serving change) reads whole instead of walking, since the walk
keeps rows above its anchor verbatim, a walked or overlapping head still
offering a sign-in link reads whole, and a background warm that anchors
nothing in a cache with a raw retained count reads whole once so a remote
clear is still read as a shrink. The refreshSlotBound over-ceiling test is
billygerhard added a commit to billygerhard/KiroCrew that referenced this pull request Oct 3, 2026
Rebased onto kirodotdev#13236 and kirodotdev#16005, which bound refreshSlot, switchSlot and
warmSlotCache with a clamp plus an older-walking window. This keeps that walk
on all three and adds only what a kept head needs on top of it: a marked view
or cache (a serving change) reads whole instead of walking, since the walk
keeps rows above its anchor verbatim, a walked or overlapping head still
offering a sign-in link reads whole, and a background warm that anchors
nothing in a cache with a raw retained count reads whole once so a remote
clear is still read as a shrink. The refreshSlotBound over-ceiling test is
billygerhard added a commit to billygerhard/KiroCrew that referenced this pull request Oct 3, 2026
Rebased onto kirodotdev#13236 and kirodotdev#16005, which bound refreshSlot, switchSlot and
warmSlotCache with a clamp plus an older-walking window. This keeps that walk
on all three and adds only what a kept head needs on top of it: a marked view
or cache (a serving change) reads whole instead of walking, since the walk
keeps rows above its anchor verbatim, a walked or overlapping head still
offering a sign-in link reads whole, and a background warm that anchors
nothing in a cache with a raw retained count reads whole once so a remote
clear is still read as a shrink. The refreshSlotBound over-ceiling test is
billygerhard added a commit to billygerhard/KiroCrew that referenced this pull request Oct 3, 2026
Rebased onto kirodotdev#13236 and kirodotdev#16005, which bound refreshSlot, switchSlot and
warmSlotCache with a clamp plus an older-walking window. This keeps that walk
on all three and adds only what a kept head needs on top of it: a marked view
or cache (a serving change) reads whole instead of walking, since the walk
keeps rows above its anchor verbatim, a walked or overlapping head still
offering a sign-in link reads whole, and a background warm that anchors
nothing in a cache with a raw retained count reads whole once so a remote
clear is still read as a shrink. The refreshSlotBound over-ceiling test is
billygerhard added a commit to billygerhard/KiroCrew that referenced this pull request Oct 5, 2026
Rebased onto kirodotdev#13236 and kirodotdev#16005, which bound refreshSlot, switchSlot and
warmSlotCache with a clamp plus an older-walking window. This keeps that walk
on all three and adds only what a kept head needs on top of it: a marked view
or cache (a serving change) reads whole instead of walking, since the walk
keeps rows above its anchor verbatim, a walked or overlapping head still
offering a sign-in link reads whole, and a background warm that anchors
nothing in a cache with a raw retained count reads whole once so a remote
clear is still read as a shrink. The refreshSlotBound over-ceiling test is
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants