Plan and provision a database

You can use a data contract to generate database tables and constraints, review the SQL, and then deploy it to a destination database. Packata provides these operations through the .NET library. A CLI provisioning command is planned; packata provision is not implemented.

Prepare the contract and destination

Load your descriptor and map it to Packata’s shared contract model. The .NET Quickstart shows how to load a Data Package descriptor; the .NET reference also lists the ODCS entry points.

Configure a DubUrlProvisioner for the destination database. You will need a connection service, a provider for that database, and any credentials required by your application. Follow the database connection guide to set these up. The examples below assume that provisioner is configured and contract contains the mapped contract.

Check which operations the provider supports

Before choosing an operation, inspect its capabilities:

var support = provisioner.GetOperationSupport();
Console.WriteLine($"Schema deployment: {support.SchemaDeployment}");
Console.WriteLine($"Bulk loading: {support.BulkLoading}");

Schema deployment and bulk loading are separate capabilities. A provider may be able to create tables without offering a bulk-loading adapter. Reading from a database is another independent capability, so a working reader alone does not establish deployment support.

The preflight details explain the supported dialects and how capabilities are determined.

Generate and review the SQL

Create a plan before changing the destination:

var plan = provisioner.PlanSchema(contract);
Console.WriteLine(plan.Script);
foreach (var diagnostic in plan.Diagnostics)
    Console.Error.WriteLine(diagnostic);

PlanSchema generates the SQL without deploying it. Read the script and diagnostics and confirm that the destination is correct before proceeding.

The plan uses mapped physical table and column names, supported physical types, and schema qualification. Depending on the dialect, it can include composite primary keys, foreign keys, range, length, pattern and enum constraints, and comments.

Diagnostics identify unresolved names and unsupported types, constraints or dialect features. A relationship in the contract produces foreign-key SQL only when it represents a supported foreign key; other relationships remain in the model.

Deploy the reviewed plan

When you are ready to apply the script, pass the plan you reviewed:

provisioner.DeploySchema(plan);

Packata executes that plan’s script without generating it again. If your application does not need a separate review step, DeploySchema(contract, options) generates and deploys directly.

Capability checks can reject an unsupported operation before SQL runs. They do not establish that the database account has permission, that deployment will succeed, or that a failure can be rolled back transactionally.

Load data after deployment

Use LoadDataAsync to load data into the destination, or ExecuteAsync to combine schema deployment and loading. The combined operation checks deployment support, bulk-loading support and reader configuration before running SQL or reading source data. Configure the source readers as described in the reader guide.

For a complete runnable example, see examples/DatabaseConnections in the repository. It creates a DuckDB table and reads it through both ODCS and Data Package contracts. The database guide provides the run command and explains connector setup and resource ownership.

CLI schema inference discovers metadata from an existing database, and JSON Schema export describes a record. Neither command creates database tables.