aax
무료Optional AAX support for Pulp, including developer-supplied Avid SDK setup, CMake enablement, DigiShell/AAX Validator workflows, and local AAX builds on macOS or Windows.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
Use Pulp's locked mise toolchain consistently across macOS, Linux, and Windows. Apply this when building Pulp, creating a disposable Windows VM, investigating toolchain drift, or changing the canonical toolchain manifest. Keep mise opt-in at shell entry, fail closed on lock drift, and distinguish pinned toolchain evidence from a build made with host-installed tools.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
설치 방법 보기설치하기 전에 에이전트에게 전달되는 지시 내용을 확인할 수 있습니다.
Use this skill whenever a task depends on build-tool versions. The canonical
input is tools/toolchain/manifest.toml; .mise.toolchain.toml and
mise.toolchain.lock are generated outputs. Never hand-edit generated files.
From the repository root, verify the checkout and install only the locked tools:
mise --version
mise trust
MISE_ENV=toolchain mise install --locked
MISE_ENV=toolchain mise exec -- python3 tools/toolchain/check_env.py
MISE_ENV=toolchain mise exec -- pulp build
Use MISE_ENV=toolchain mise exec -- <command> for one-off commands so the
versioned PATH is explicit in logs. On Windows, use the same commands from
PowerShell (replace python3 with python only when the pinned interpreter's
launcher requires it). Do not use mise install without --locked for a Pulp
receipt.
A pinned-toolchain claim requires all of the following:
[mise].version in the manifest;mise install --locked on the guest's platform;tools/toolchain/check_env.py with no unexpected drift; andIf any item is absent, label the result host-toolchain evidence. Do not infer pinned versions from a generated config that was not actually used.
A fleet host, an ssh session or a launchd job has neither the pinned mise's
directory nor the rustup proxies (~/.cargo/bin) on its PATH, so a plain
check_env.py there reports an installed Rust as missing. Rollout steps run
the one command that judges them:
MISE_ENV=toolchain mise exec -- python3 tools/toolchain/check_env.py --fleet-path --require-tools
--fleet-path also searches those directories; --require-tools exits 1 on
any host-tool difference whatever the manifest's mode, while only reporting the
npm trees each checkout installs for itself (pulp deps resolve).
tools/toolchain/fleet_report.py runs the check over ssh on several hosts and
prints one drift table; it installs nothing.
Read the versions a build actually used from its CMakeCache and configure log
(CMAKE_COMMAND, CMAKE_MAKE_PROGRAM, Found Python3), not from the profile.
On macOS, CMake's FindPython prefers a Python.framework install over PATH; the
configure of a Pulp checkout sets Python3_FIND_FRAMEWORK=LAST when
MISE_ENV=toolchain is set so the pinned interpreter wins. A build directory
configured before opting in keeps its cached interpreter: configure a fresh one.
.shipyard/vm-image.mise.toml bakes a candidate image,
pulp-build-runner-mise-candidate, on top of the runner golden with
tools/ci/mise-golden-layer.sh (run by tart-provision.sh manifest). Its
[mise] table is generated from the manifest; regenerate it with
generate.py, never by hand. The layer refuses a host or guest mise binary
whose sha256 or --version differs from the pin, and toolchain files that
differ from the checkout's commit, so bake from a clean main. It installs
with mise install --locked, turns rustup self-update off and installs the
manifest's Rust by name, then writes a receipt the guest itself reports
(on the host: ~/.cache/pulp-ci/golden-receipts/<vm>-mise.json).
The candidate changes no lane: promoting it to :latest or routing a lane to
it is a separate, reviewed step.
Some tools are not locked on every platform, by declaration in the manifest's
unlocked table: Tart is macOS-only, and conda-forge publishes no
windows-arm64 clang-format, so windows-arm64 keeps its own clang-format. The
drift test refuses any other missing platform row; never "fix" an exception by
hand-editing the lock.
MISE_ENV=toolchain opt-in at shell entry. It prevents a checkout from
rewriting a developer's PATH or installing tools unexpectedly.setup.sh may still run on
hosts without mise while their reviewed provisioning migration is incomplete..mise.toml. Put those in the VM recipe or host configuration.pulp build/pulp dev or the governed CI wrapper; mise controls
tool versions, not build parallelism or runner admission.To change a tool, edit the manifest on a branch and regenerate with the exact mise binary named by the manifest:
python3 tools/toolchain/update.py cmake --to 4.3.4 --mise /path/to/mise
python3 tools/toolchain/generate.py --check
Run the toolchain self-tests and dependency audit before publishing. A pin
change must update the manifest, generated mise files, lock checksum, and
DEPENDENCIES.md/licensing documentation when the dependency inventory changes.
Do not claim a pin change is complete until the generated-file check is clean.
Moving mise itself ([mise] version, source_commit, [mise.binary]) is a
source build of the release tag, not an update.py run. Before claiming "no
relock", run generate.py --lock --mise <new binary> in a throwaway clone:
it deletes and rewrites the lock, so a fresh lock mtime plus a clean
git status proves the new mise reproduces it byte for byte. --check alone
never runs mise and cannot show that. On the fleet, the tartci profile's
[mise] table refuses the host install when its binary is missing, so the new
binary goes onto a host before that host's table changes, and the previous
binary stays installed as the rollback. A source build at a new path is a new
program to LuLu on every host; flipping the pin needs no network (the tools are
already installed), so pre-seed rather than asking for firewall rules.
New mise releases surface in the weekly "Dependency drift: weekly audit"
issue (deps-drift-check.yml): tools/deps/manifest.json tracks mise as a
git-tag upstream, and a test refuses it unless its version and tag match
[mise].version. A newer release there starts the evidence path above; it
never moves a host by itself.
For an ephemeral ARM64 or x64 Windows guest, copy only the locked config and use a guest-local mise cache. Keep the guest isolated from the persistent interactive VM, and destroy the overlay after the receipt is copied out. A successful configure or SDK build is useful evidence, but it is not plugin GUI or DAW proof; retain that distinction in the status report.
아직 리뷰가 없습니다. 사용해 본 소감을 남겨 주세요.
로그인 없이 쓸 수 있습니다. 사용 후기, 질문, 작동하지 않은 환경 등을 편하게 남겨 주세요.
아직 댓글이 없습니다. 첫 댓글을 남겨 보세요.
Optional AAX support for Pulp, including developer-supplied Avid SDK setup, CMake enablement, DigiShell/AAX Validator workflows, and local AAX builds on macOS or Windows.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
Configure, implement, and test Pulp's optional desktop Ableton Link tempo-sync adapter while preserving the developer-supplied SDK, licensing, realtime, latency-compensation, and no-install boundaries.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
Maintain Pulp's installed design-time agent capability manifest and public-surface ledger. Use when adding, removing, renaming, or materially changing public audio, MIDI, signal, timebase, or sequence APIs; registering a new algorithm for generators; changing capability support or deprecation state; or repairing agent-capabilities freshness, schema, fingerprint, tombstone, or installed-SDK tests.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
Android platform development for Pulp — NDK cross-compilation, Oboe audio, Dawn/Skia GPU rendering, JNI bridge, touch interaction, emulator workflows, and end-to-end smoke validation. Covers build, deploy, debug, and the gotchas discovered during bringup.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
Optional ARA support for Pulp, including developer-supplied ARA SDK setup, CMake enablement, adapter companion APIs, validation, and ARA-aware plugin implementation guidance.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
The measurement surface for ALL Pulp DSP and audio-pipeline work — read it BEFORE writing or gating DSP, not only when something already sounds wrong. Covers the C++ harness (signal generators, metrics, assertions, RenderScenario, contracts), the offline Audio Doctor (magnitude/frequency response, THD/THD+N, phase/group delay), and their Python sibling the Audio Quality Lab (tools/audio/quality-lab — null residual + alignment, LTAS log-spectral distance, spectral flux/centroid, HNR, Theil-Sen drift slope, Kaiser-sinc resampling, license-guarded corpus, regression-net ratchet). TRIGGER on AUTHORING work — "build/design an oscillator/filter/synth/effect", "add a DSP module", "what should the acceptance gate be", "how do I measure aliasing / anti-aliasing / alias floor", "null against a reference", "is this DSP correct", "choose a tolerance", "golden/regression corpus for audio", "measure drift or jitter", "A/B two renders" — AND on DEBUGGING work — "is there sound / no audio / I hear nothing", "does this filter/compressor/synth/delay produce the right signal", "prove the DSP / prove the contract", "measure the frequency response", "what's the THD / is it distorting", "what's the group delay / phase response / measured latency", "magnitude response curve", "render a test tone and assert", "audio regression", "64-frame works but 128 is silent", "sample-rate change pitch-shifted it", "describe what's in this buffer", "audio doctor", "compare before/after a DSP refactor". Reach for this BEFORE hand-rolling any FFT, null test, alias measurement, pitch tracker, or golden-render script — most of it already exists in one of the two lanes. Test/tool layer over HeadlessHost — deterministic, no audio device, no speakers. Off the realtime thread entirely.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.