Azure Functions Host

The JSON schemas in this repository are copies of the official schemas published to the JSON Schema Store. These schemas are used by Azure Functions tooling, such as Visual Studio, to provide validation and IntelliSense for configuration files.

Important Maintenance Notes:

• host.json schema: This schema is no longer maintained in this repository; it is now maintained in the Azure/Azure-Functions repository.
• Making changes: If you need to update a schema, you must submit a Pull Request to the Schema Store Repository rather than this repository. Changes made here are merely copies.

Private Site Extensions (PSE) are not generated during the standard build process. To generate one, navigate to the publish output directory ({repo_root}/out/pub/WebJobs.Script.SiteExtension/{config}_win) and run the New-PrivateSiteExtension.ps1 script.

By default, it generates a zipped x64 PSE.

# Generates a zipped x64 PSE by default
./New-PrivateSiteExtension.ps1

# To generate x86 / 32bit:
./New-PrivateSiteExtension.ps1 -Bitness x86

# Can skip zipping the extension:
./New-PrivateSiteExtension.ps1 -NoZip

There are two ways to produce a compressed (.zip) site extension:

1. Via dotnet publish: Add the -p:ZipAfterPublish=true flag to your publish command. The resulting zip will be located at {repo_root}/out/pkg/{config}.
2. Via PowerShell script: If you have already published the unzipped files to {repo_root}/out/pub/WebJobs.Script.SiteExtension/{config}_win, you can run Compress-SiteExtension.ps1. This method allows you to insert JIT trace files during the compression process.

# Produces the .zip site extension by default
./Compress-SiteExtension.ps1

# Produce the zip, inserting JIT trace files beforehand.
./Compress-SiteExtension.ps1 -JitTrace "path/to/file.jittrace", "path/to/file2.jittrace"

The Site Extension can be built and published using MSBuild commands. You can perform these steps together in a single command or separately for more granular control.

Note: When building separately, you must set PublishReadyToRun=true (this requirement is fixed in the .NET 9 SDK).

# Together
dotnet publish -c {config}

# Separately
dotnet restore -p:PublishReadyToRun=true
dotnet build -c {config} --no-restore
dotnet publish -c {config} --no-build

To execute the performance benchmarks for the WebJobs.Script components, use the dotnet run command from the solution root. You can either use the interactive prompt to select specific benchmarks or pass a filter directly via the command line.

Interactive Mode

Running the base command will present a list of discovered benchmarks. You can select them by entering their index number (e.g., 0), their caption (e.g., AuthUtilityBenchmarks), or multiple benchmarks separated by spaces (e.g., 1 2 3).

Filtered Mode

To bypass the interactive prompt and run specific benchmarks immediately, use the --filter flag with a glob pattern (e.g., *BenchmarkName*).

For Java workers, use the protobuf-maven-plugin. Configure the protoSourceRoot in your pom.xml to point to the directory containing the .proto files within the subtree.

Configuration:

<protoSourceRoot>${basedir}/<path to this repo>/azure-functions-language-worker-protobuf/src/proto</protoSourceRoot>

Use the protobufjs npm package to generate runtime code from the protobuf definition. You can generate CommonJS JavaScript modules or static TypeScript definitions.

Generate JavaScript (CommonJS): Use pbjs with the -t json-module flag.

Generate TypeScript: Use pbjs with the -t static-module flag to create a static JS file, then use pbts to generate the corresponding .d.ts declaration file.

To build the Python implementation, install the package in editable mode with dev dependencies and run the setup build command.

python -m pip install -e .[dev] -U
python setup.py build