Repository navigation
fix(website): re-measure the App chunk's bundle-size ceiling - #33669
Pearcekieser wants to merge 1 commit into
Conversation
First Principles Review (Opus 5.5) — ⏭️ skippedRevision |
UX Review (Opus 5.5) — ✅ PASSUX-level review of UX-Verdict: PASS Nothing to check. The diff only changes a build script's bundle-size ceiling, with no user-visible control, so the missing blind read leaves no evidence gap. [UX-REVIEWED] 9bccbe0 Description read[DESCRIPTION-READ] ad87bb041295e03e748ad74b63eaad630ff2d405312acb5dfcaad5b3cde46bcd 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 |
Design Review (Opus 5.5) — ✅ PASSDesign-level review of Design-Verdict: PASS Nothing blocks. This is a budget re-measure that follows the convention the comment block already sets, it renders nothing new, and a revert undoes it cleanly. Notes
[DESIGN-REVIEWED] 9bccbe0 Description read[DESCRIPTION-READ] 4b22a03e4b08eec9487119223eb2940f21a082f859338f4aa99d40811ea190c9 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 |
Opus 5.5 Review — ✅ no blocking findingsReviewed Verdict parsed from the review's SHA-scoped output markers for commit False positive or not applicable? A repository writer can comment: |
GPT 6.1 Review — ✅ no blocking findingsGPT 6.1 completed its review of This comment is updated in place on each push. Review detailsNo findings. False positive or not applicable? A repository writer can comment: |
Problem / Motivation
Goal: The per-chunk Bundle Size Gate passes on main again, so PRs that touch
website/stop failing it for main's own growth.On main at f08ae0d, the analyze build emits the
Appchunk at 2,253,662 B. Its ceiling is 2200 KB (2,252,800 B), so main alone is 862 B over, and every PR that toucheswebsite/inherits the red check on its merge ref. The Main Ratchet Audit tracks the same lane in #32956.Why it matters
A gate that fails on main blocks unrelated PRs and teaches authors to rerun or ignore it, which hides a real regression when one comes. The gate exists to catch a new library or surface landing in the always-loaded chunk, not 29 days of spread-out growth.
Not a goal
Appchunk smaller. The growth is spread across the always-loaded shell and chat route, and the route-only pages are already lazy, so no singleimport()boundary moves it out.What changed (motivation → approach → change)
CHUNK_BUDGETSinwebsite/scripts/check-bundle-size.mjssetAppto 2200 KB from a 2,144,668 B measurement on 2026-10-11 (#19089), about 5% headroom. Since then, first-party growth added 108,994 B and 77 modules to that chunk, so the headroom is spent and main's tip is over.The ceiling is re-measured on main's tip with the same ~5% convention the comment block prescribes: 2,253,662 B * 1.05 = 2,366,346 B, rounded up to the table's whole-KB unit, which is 2311 KB (2,366,464 B, 112,802 B or 5.01% of headroom). A new paragraph in the comment block records the measurement, the commit, the module count and why no lazy boundary applies, in the same form as the earlier re-measurements.
Backwards compatibility
Compatible. Only the
Appceiling changes. No product code changes, and every other budget stays as it is.Tests
The gate itself is the test. On a clean worktree at f08ae0d:
node --max-old-space-size=6144 ./node_modules/vite/bin/vite.js build --mode analyze, thennode scripts/check-bundle-size.mjs, fails with "exceeds its 2.15 MB budget by 862 B (chunk 'App')" before this change.Proof: not provable -- frontend test (prove.py runs Python only)
Manual verification
N/A. The gate run above is the verification.
Screenshots / video
N/A. No user-visible change.
Related Issues
The Main Ratchet Audit reports the same
bundle-ceilinglane in #32956. The last re-measurement was #19089.Sibling sweep
Swept: node scripts/check-bundle-size.mjs on main @ f08ae0d
Siblings: none -- the gate run lists App as the only chunk over its budget on main
Pattern harvest
Not generalizable: a per-chunk ceiling with a fixed headroom is spent by ordinary growth by design, and the comment block already documents the re-measure step.
Checklist
Contribution License Agreement
Contributions are accepted under the project's Apache-2.0 license.
Fixes #33668