Quirlv0.7.0
ResearchBenchmark archive

Quirl 0.7.0 native release evidence

Canonical Quirl project documentation synced from docs/benchmarks/release-v0.7.0.md.

Recorded 2026-11-12 for the published v0.7.0 release. All four native artifacts passed the enforcing release benchmark on their first attempt, with 101 successful PTY samples each. These results measure the exact published executables on the hosted runners below; they do not establish universal latency guarantees.

Release identity

  • Candidate and resolved tag commit: bbfc8259c7cdea2044b1d804f76a388aacb39b31.
  • Native build and performance run: 43185798988.
  • GitHub release ID: 442712045; published at 2026-11-12T09:12:19Z as a nondraft, non-prerelease, immutable release.
  • Release manifest SHA-256: d27319457af49961e6c7f3cd12578d6005d2b151d2112a7f527e1f74f3e3a7f8. Release notes SHA-256: 4be4d3d2c25429c2aa94c2ca518914ecc2ba5ec9bac490d1068f5c5953f37899.
  • Every entry in the bundle's SHA256SUMS verified before publication was approved. All 15 published assets were downloaded afterwards and matched the reviewed bundle byte for byte, and the four formula hashes in tap PR 48 matched the archives.

Each report records the candidate as its clean source revision, with artifact digest, profile, source, and harness identity checks passing.

Native results

All values below are milliseconds except executable bytes. Limits were startup P50 ≤25 ms, first-prompt paint P95 ≤21 ms, and keystroke-to-frame P95 ≤8 ms. Each row completed 101 of 101 requested PTY samples with no recorded sample failure.

TargetExecutable bytesStartup P50First-prompt P95Keystroke P95Result
Linux ARM6412,820,9685.9688364.4379390.514188Pass
Linux x86_6414,982,8406.2765554.6657880.505195Pass
macOS ARM6410,894,76813.60725015.7663751.248958Pass
macOS x86_6413,158,76812.52341012.1315671.273134Pass

Size follows ADR 0036: there is no default executable-size ceiling, and every size above the advisory 8 MiB threshold retained its warning.

The macOS x86_64 startup P50 fell from 19.51 ms in 0.6.1 to 12.52 ms, close to where the macOS ARM64 runner measured, and the macOS ARM64 first-prompt P95 fell from 17.86 ms to 15.77 ms. The source change between the releases is the variable import after source, ., eval, and POSIX islands, which is not on the startup path, so the reports alone do not attribute either difference to Quirl. The Linux x86_64 runner again reported a different CPU than for the previous release.

Runners and toolchain

TargetGitHub runner labelCPU reported by harnessLogical CPUsOperating system
Linux ARM64ubuntu-24.04-armunknown4Ubuntu 24.04.5 LTS
Linux x86_64ubuntu-24.04AMD EPYC 9V74 80-Core Processor4Ubuntu 24.04.5 LTS
macOS ARM64macos-15Apple M1 (Virtual)3macOS 15.8.2 (24H232)
macOS x86_64macos-15-intelIntel(R) Core(TM) i7-8700B CPU @ 3.20GHz4macOS 15.7.9 (24G830)

All builds used Rust 1.99.0 (b940084d7 2026-09-28) and LLVM 23.1.1 with the official release profile: optimization z, fat LTO, one codegen unit, stripped symbols, and panic = "unwind".

Exact report files

These are byte-for-byte copies of the reports uploaded by the passing native jobs.

Target and exact reportJSON SHA-256Native job
aarch64-unknown-linux-gnuabf02141881544b7d8f356a9fe800b72940499e767677930fc29e9851f3eabfb131076614519
x86_64-unknown-linux-gnu8fb54a12dad19e3f7788c76290b028654deb0f025f7c825b7c602ca9342de535131076614353
aarch64-apple-darwina759a74e6404212ede8192859a5f426548985061177c88ebbebfa99d4d729494131076614281
x86_64-apple-darwin137a34bbf99a4edb1413fa1f56033825e15f39bf2a21c1965eb87200e87248ae131076614339

Published executable hashes

Each digest below is the bin/quirl executable inside the published archive. It equals the digest the passing report measured.

TargetExecutable SHA-256
aarch64-unknown-linux-gnu5b1120232387dd55341ea2a36e40016311208ff6032d0ef71084d00a27dc2ae6
x86_64-unknown-linux-gnucf076095c8b77416f1041adccab783d394c626cbd9d72f47488d6da25dda17b4
aarch64-apple-darwinc2ddc1772bf4b8540426ff8857558078681f2bbf6a7be2c9f616e94b30c99a6c
x86_64-apple-darwin4379340f2d64e1f0d16d23f06e8e24b0bba18999c9857877db1aee4ffaa7bf7d

Method and limitations

The PTY harness uses a private, initially empty home, configuration, history, catalog, project database, and XDG directories shared across its samples. Each sample starts a fresh native process in a 120×40 terminal, answers cursor-position queries, reconstructs the screen, checks the prompt and exact binary identity, then proves editability. Percentiles use nearest-rank selection. Hosted scheduling, virtualization, CPU frequency, other machine load, and filesystem caches are not controlled beyond the documented indexing and settling steps. The reconstructed PTY frame does not measure physical terminal-emulator scheduling or display. This automated evidence does not cover every terminal, visual accessibility scenario, clipboard, IME, or sustained human session, and no human checklist item is inferred from it.

The website recording, its poster, and the README and website screenshots were recorded from the published macOS ARM64 executable of this release, SHA-256 c2ddc1772bf4b8540426ff8857558078681f2bbf6a7be2c9f616e94b30c99a6c, through scripts/record-demo.sh; their provenance records every digest.

On this page