Installation¶
Weaver needs Python 3.11 or later. Install it into the Python environment you use for the project, ideally a virtual environment:
Install Weaver¶
The package provides the weaver command and the weaver Python library. weaver --help lists the commands.
Sign in for the operation¶
The weaver command signs in with the first credential that works: a configured service principal, then an existing Azure CLI sign-in, then interactive browser sign-in. On a desktop, either sign in with the Azure CLI first or let the first command open a browser:
The identity needs the workspace rights each task lists. Tasks state their own requirements, such as creating Warehouses, publishing an Environment or refreshing a semantic model, where you need them. Unattended execution covers service principals and --non-interactive for jobs.
Model and Report access¶
Semantic models and Reports can run in an existing, suitably licensed Power BI workspace without a Fabric Warehouse, Lakehouse or Catalogue. Fabric data workloads need Fabric capacity. See the two operating modes and model and Report requirements.
Publish the Fabric runtime¶
Lakehouse Python and Spark SQL run in Fabric Spark, so the Fabric Environment needs the same Weaver version you installed locally. The first Lakehouse task publishes it with one command before its first Build. Warehouse T-SQL and Power BI-only work do not need it.
Next¶
Start with a first task. The Warehouse Table and View is the shortest first result.