27a97ee619 vs head 7616ba222d
iOS snapshot benchmark evidence
Orphan branch evidence/ios-snapshot of callstack/agent-device. It holds the raw
pnpm bench:ios-snapshot results measured at repository commit
71fb2483f30d90e615e949601c836aeebbf450c5 on bench-golden-v2 (iPhone 17 Pro, iOS 27.0).
The files are measurement output, not fixtures; the harness, the schema
(scripts/ios-snapshot-benchmark/raw-result.schema.v1.json), and the adjacent Markdown
summaries live on main under scripts/ios-snapshot-benchmark/.
| File | sha256 |
|---|---|
ios-snapshot-cold-local-71fb2483f.json |
532a83247bfbf8ee47039f80ac429f067c84679e92c781768c1044da1ae6e9bf |
ios-snapshot-warm-relaunch-local-71fb2483f.json |
6d299e8baec69662dca2c1ad8f1348e4361d5afaa781080e9a6b9b3dac362cbf |
ios-snapshot-proxy-71fb2483f.json |
b11b7a07be9e4dcf003f3af66943682a6733c6f21f5f43d3d9e88b3fb37b51a7 |
Simulator accessibility bridge decision evidence
These compressed artifacts support the Node-direct Simulator bridge decision measured at
repository commit 268a90275e7a30419e581336b6d85eff680a2eb6. The broad corpus retains the
September 1 measurements; the targeted and corrected artifacts contain the clean rerun and
maintainer-corrected evaluation.
| File | sha256 |
|---|---|
ios-simulator-ax-bridge-broad-268a90275.json.gz |
309f974b1dcb90768548a189f6af58b493b5d7b9d56a5bfad060d4335139eb7b |
ios-simulator-ax-bridge-targeted-268a90275.json.gz |
fa2e01dcb5e2a0229a6836f1a4187169445347dd4c0a42bcb5a29022538c70b2 |
ios-simulator-ax-bridge-corrected-268a90275.json.gz |
4d70596a39153e6104e37a790622450d967f61b2244745915f39462514ba3bed |
Corrected relaunch corpus
These artifacts were captured at repository commit
636b1deac98ab88cc8e0e1ed894b5719d8a6c83f. They replace the targeted decision
artifact above with 20 Node-direct relaunch samples on each of six representative
screens. Every timed sample is paired with independently observed readiness for the
expected application process generation and screen anchor. The supplied idb guest
binary was verified against its pinned SHA-256 before capture.
| File | sha256 |
|---|---|
ios-simulator-ax-bridge-targeted-636b1deac.json.gz |
092d3deab3753c1b7a0d230d9e54f703e5fedb87ffc0974f082541ba9b4e687d |
ios-simulator-ax-bridge-corrected-636b1deac.json.gz |
a20039b38d4da65fed65518a3214153afa0174faaf6fb38c3950a3bd1362870d |
Post-rebase final corpus
The final artifacts were recaptured at repository commit
44995806ea3be09f3c48ceac50ac3cab18462c35 after rebasing the pull request onto
941ca0e7e08f50960dc90d3fdafa572355b030a0. They repeat the same bounded corpus
and supersede the preceding targeted artifacts for the pull-request decision.
| File | sha256 |
|---|---|
ios-simulator-ax-bridge-targeted-44995806ea.json.gz |
3440d066cb7eea33c4715fece838b5185c5d209694b097e54f5536d48d4984ad |
ios-simulator-ax-bridge-corrected-44995806ea.json.gz |
0a34a84402e85154e177adef5122101b26bfd04d74714a0a9b6f0795270edc41 |
Production routing acceptance (#2279)
PR #2279 acceptance evidence contains the matched six-screen
warm corpus at ae26f7afc064c75c65c71e677d32d77cdf2f9709 against
7a2d48d160aaacb582c6f4cde98b6af7531bf7af, public responses, foreground fallback
proof, regression records, and checksums. The immutable tag is
evidence/ios-snapshot/ae26f7afc0. Timings are bounded observations under
uncontrolled host load, not a general performance guarantee.
Fetch into a checkout
git fetch origin evidence/ios-snapshot
for f in ios-snapshot-cold-local-71fb2483f.json \
ios-snapshot-warm-relaunch-local-71fb2483f.json \
ios-snapshot-proxy-71fb2483f.json; do
git show FETCH_HEAD:$f > scripts/ios-snapshot-benchmark/evidence/$f
done
shasum -a 256 scripts/ios-snapshot-benchmark/evidence/*.json
pnpm bench:ios-snapshot:evidence
Add a new corpus
Commit new raw results on top of this branch with the message
evidence(ios-snapshot): benchmark results measured at <commit> and extend the table
above. Never rewrite history here: main cites these hashes.