feat(ato): Surface Activation v2 + vite production build lane - #1231
Merged
Conversation
…builder A builder's scratch directory is not a durable record, and a number in a PR comment is only as good as the log behind it. This is the raw 10/10 run that gated the merge, plus the steps to regenerate it — an unreproducible measurement is an anecdote. Records the caveats alongside the result rather than only the result: ten runs is not a p99 sample; the E2E supplies its own AcceptanceConfig, so what was measured is the WORK and not the production budget path; and the production builder held no live job during the window, so contention classification A rests on the lock namespaces being structurally separate plus no interference observed, not on a contended run. Carries the two product findings the run produced (ato#1157 vsock leak, ato#1160 unbounded capture retry) and the warning that run-uffd-kvm-smokes.sh is unusable on a live runner. Stored as .txt because .gitignore excludes *.log — the point is that this file is tracked.
…erver Community capsules inferred from a package.json whose scripts declare plain `vite build` + `vite preview` now launch `sh -lc "<pm> run build && <pm> run preview -- --host 0.0.0.0 --port 8000 --strictPort"` instead of the dev server. Dev serving shipped the unbundled module graph through the app proxy — measured for drawdb: 448 requests / 115 MB before first paint (the 'ready but blank for ~30s' preview); swagger-editor's graph is 542 modules and never finishes booting. The build runs once at snapshot capture (restores resume the serving process), and the pinned v1 lane grants build-then-serve commands the capped 600 s boot budget for that capture. Compound build scripts or a missing preview script keep the dev lane (fail-closed, unchanged behavior). Claude-Session: https://claude.ai/code/session_015MNKB1dibFqVffp5UQcg8V
…at boot The first cut launched `sh -lc "npm run build && npm run preview …"` — but the v1 guest rootfs is mounted read-only (only tool-owned cache paths are tmpfs-redirected), so `vite build` writing dist/ dies on EROFS at boot and the capture burns its whole readiness budget (observed twice on the staging builder: 600 s spent, guest never healthy; the same build in an equivalent 2-cpu/6 GiB container takes 11 s). The build now runs as part of image assembly: build_v1 chains `<pm> run build` after dependency install when the launch is `<pm> run preview …` and package.json declares a plain `… vite build` script (vite_production_prebuild_cmd — evidence-based and fail-closed, same doctrine as base_image_and_install). The inferred launch becomes a plain `<pm> run preview -- --host 0.0.0.0 --port 8000 --strictPort` argv with the standard 60 s readiness, and dist/ ships read-only in the identity-bearing guest tree. The now-unneeded shell-compound seal arm and capture boot-budget special case are reverted. Claude-Session: https://claude.ai/code/session_015MNKB1dibFqVffp5UQcg8V
The 9 rootfs_builder tests that execute a real `sh` builder script (npm-ci classification, stderr redaction/truncation) or rely on unix staging-dir semantics fail on Windows CI, where sh isn't the shell — same platform gating as the render-budget tests already on main. Claude-Session: https://claude.ai/code/session_015MNKB1dibFqVffp5UQcg8V
Koh0920
added a commit
that referenced
this pull request
Aug 2, 2026
feat(ato): Surface Activation v2 + vite production build lane
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
1 participant
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.
Merges the staging-verified Surface Activation v2 arc (ready ack v2 public-surface activation + the vite production-build lane that fixed the 30s ready→visible gap; drawdb-3 FCP 8.1s→1.46s measured on staging) plus the ato#1155 hardware evidence doc.
Already deployed and verified on the staging builder (Sugamo) and exercised by every staging intake since. This PR lands the source on main; it is also the base of the ADR-016 CPU-entitlement stack (#1225→#1226→#1228).
Supersedes #1229 (identical patch-ids for both vite commits — verified with git patch-id --stable).
🤖 Generated with Claude Code