Skip to content

04-validate-and-install.md

Latest commit

 

History

History
196 lines (134 loc) · 5.68 KB

04-validate-and-install.md

File metadata and controls

196 lines (134 loc) · 5.68 KB

04: Validate and Install

Concepts covered:

  • Package validation process and schema verification
  • Dependency resolution and installation orchestration
  • Package installation flows and error handling

Skills you will practice:

  • Validating package structure and metadata
  • Installing packages from local directories
  • Understanding validation feedback and error messages

Understanding Validation

This article covers validating your package and installing it into a Hatch environment. Now that you have implemented functionality and configured metadata, it's time to validate and test your complete package.

The validate CLI subcommand constructs a HatchPackageValidator with version='latest' and allow_local_dependencies=True, passing the environment manager's registry data into the validator. On success the command exits with code 0; on failure it exits with code 1 and prints grouped validation errors (the CLI prints category-level errors except for the special valid and metadata categories).

Step 1: Validate Package Structure

Before installing or submitting your package to the online registry, validate your package meets Hatch requirements:

hatch validate /path/to/my_package

The validation process checks:

  • Metadata schema compliance - Ensures hatch_metadata.json follows the latest schema
  • Required files presence - Verifies entry point and essential files exist
  • Field validation - Checks naming patterns, version formats, and constraints
  • Dependency specifications - Validates dependency definitions and constraints

Successful Validation

[SUCCESS] Operation completed:
  [VALIDATED] Package 'my_package'

The command will exit with status code 0 when validation succeeds.

Failed Validation

Package validation FAILED: /path/to/my_package

When validation fails the CLI prints a failure line and then — if detailed results are available — lists errors grouped by category. The CLI deliberately skips printing the valid and metadata categories in that grouped output and only prints categories which are invalid and include an errors list.

Example failure output (category grouping):

Package validation FAILED: C:\path\to\my_package

Entry Point Errors:
 - Missing required field 'entry_point'
 - Entry point 'hatch_mcp_server_entry.py' not found

Dependency Errors:
 - Dependency 'some_pkg' not found in registry

Note that metadata category details may not be printed in the grouped list by the CLI loop; important metadata validation failures may still be reported elsewhere by the validator object or surfaced in tooling that consumes the validator results.

The command will exit with status code 1 when validation fails.

Step 2: Fix Common Validation Issues

Invalid Package Name

// ❌ Invalid - contains hyphens
"name": "my_package"

// ✅ Valid - uses underscores
"name": "my_package"

Invalid Version Format

// ❌ Invalid - not semantic versioning
"version": "1.0"

// ✅ Valid - proper semantic version
"version": "1.0.0"

Missing Required Fields

Ensure all required fields are present:

  • package_schema_version
  • name
  • version
  • entry_point
  • description
  • tags
  • author
  • license

Step 3: Install Validated Package

Once validation passes, install the package:

# Set target environment
hatch env use my_dev_env

# Install the package
hatch package add /path/to/my_package

The installation process involves:

  1. Package loading - The package_loader reads and parses the package metadata
  2. Dependency resolution - The system identifies all required dependencies
  3. Installation orchestration - The dependency_installation_orchestrator coordinates installation
  4. Installer execution - Different installers handle Hatch, Python, system, and Docker dependencies

Installation Components

The installation system uses several specialized installers:

  • hatch_installer.py - Handles Hatch package dependencies
  • python_installer.py - Manages Python package installation via pip
  • system_installer.py - Handles system package installation
  • docker_installer.py - Manages Docker image dependencies
  • installer_base.py - Provides common installer interface

Step 4: Verify Installation

Confirm the package was installed successfully:

hatch package list

Output should show your package:

Packages in environment 'my_dev_env':
my_package (1.0.0)    Hatch compliant: true    source: file:///path/to/my_package    location: /env/path/my_package

Step 5: Test Package Functionality

After installation, test that your MCP server works:

# Test the package entry point
python /env/path/my_package/hatch_mcp_server_entry.py

Exercise: Create a package with intentional validation errors (invalid name, missing fields), attempt validation, fix the errors, and then successfully validate and install the package.

Solution 
# 1. Create package with errors
hatch create test-package  # Invalid name with hyphens

# Edit hatch_metadata.json to introduce errors:
# - Change name to include hyphens
# - Remove required field like "description"
# - Use invalid version format

# 2. Attempt validation (should fail)
hatch validate test-package

# 3. Fix errors:
# - Change name to "test_package"
# - Add missing required fields
# - Use proper version format like "1.0.0"

# 4. Validate again (should succeed)
hatch validate test-package

# 5. Install the corrected package
hatch env use my_dev_env
hatch package add ./test-package
hatch package list

Previous: Edit Metadata Next: Checkpoint