Skip to content

Record your screen, ship a demo. Free and open-source, GPU-accelerated, no watermarks, no subscriptions. Windows, macOS, Linux. Actively maintained.

ci: fetch win32-arm64 sharp prebuilt for arm64 packaging (non-fatal)

Christian Schmittel15aba4e4950a
diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
index 36439a64a..2f8ff31a6 100644
--- a/.github/workflows/build.yml
+++ b/.github/workflows/build.yml
@@ -74,6 +74,23 @@ jobs:
Write-Host "MSVC ARM64 build tools already present."
}
+ - name: Ensure win32-arm64 sharp binary
+ # sharp (transitive via @xenova/transformers) only fetches a host-arch
+ # prebuilt at install time, so on the x64 runner the packaged arm64 app
+ # would otherwise ship an x64 sharp.node. Best-effort fetch of the
+ # win32-arm64 prebuilt; non-fatal so it never blocks the build. Verify
+ # the first arm64 release loads sharp if auto-captions use it.
+ if: matrix.arch == 'arm64'
+ continue-on-error: true
+ shell: bash
+ run: npm_config_platform=win32 npm_config_arch=arm64 npm rebuild sharp
+
+ - name: Cache caption assets
+ uses: actions/cache@v4
+ with:
+ path: caption-assets
+ key: caption-assets-${{ runner.os }}-${{ hashFiles('scripts/fetch-caption-model.mjs') }}
+
# STT is the bundled whisper-stt-server (whisper.cpp with native DTW token
# timestamps); no model is fetched here. The binary is built by
# build-whisper-stt.yml and staged below, from the run built from this