tlaplus-cli 0.3.3

TLA+ tools: download TLC, compile custom modules, run model checker

TLA+ CLI

Command-line tool for working with TLA+ specifications and the TLC model checker.

Installation

Install system-wide via uv tool:

uv tool install tlaplus-cli

Upgrade:

uv tool upgrade tlaplus-cli

Uninstall:

uv tool uninstall tlaplus-cli

Usage

Managing TLA+ Tools

The tla tools command group allows you to download and manage multiple versions of the TLA+ toolset (TLC, SANY, TLATeX) directly from GitHub releases.

List available and installed toolset versions:

tla tools list

Install the latest toolset version:

tla tools install

[!TIP] The first version you install is automatically "pinned" as the default. Subsequent installs won't change your pin unless you manually use tla tools pin.

Install a specific toolset version:

tla tools install v1.8.0

Pin a specific version to be used by default:

tla tools pin v1.8.0

Upgrade the pinned version (or a specific version) to a newer commit:

tla tools upgrade

[!NOTE] If the target version to upgrade is not yet installed locally, the CLI will automatically download it.

Show the absolute path to the pinned version's tla2tools.jar:

tla tools path

Or for a specific version:

tla tools path v1.8.0

Example output:

/home/bob/.cache/tla/tools/v1.8.0-5a47802/tla2tools.jar

Show the toolset versions directory and all installed version directories:

tla tools dir

Example output:

/home/bob/.cache/tla/tools
  v1.7.0-abc1234
  v1.8.0-5a47802

Uninstall a specific version (or use 'default' to remove legacy jars):

tla tools uninstall v1.8.0

[!TIP] Use --all to remove all installed tags for a specific version name without interactive prompts.

[!NOTE] If you uninstall the currently pinned version, the CLI will automatically "fall back" to the next best installed version (ranked by semver, then release date).

Run TLC

Run the TLC model checker on a specification. This uses the currently pinned toolset version.

tla tlc <spec_name>

For example (runs queue.tla):

tla tlc queue

Project-Aware Execution

The CLI intelligently resolves the specification file and its project structure. It will check the following locations:

1. <spec_name> (if it's a direct path to a file, e.g., models/my_model.tla)
2. <spec_name>.tla
3. spec/<spec_name>.tla (checks an inner spec/ folder relative to the spec's location)

Furthermore, it automatically discovers the project root by looking for modules/, classes/, or lib/ directories adjacent to the spec file or its parent. When a project root is found:

• The project's classes/ directory is added to the Java classpath.
• Any *.jar files found in the project's lib/ directory are added to the Java classpath.
• The -DTLA-Library system property is set to the project's modules/ directory, allowing TLC to find your custom Java overrides.

To check the currently pinned tla2tools.jar path and its TLC version:

tla tlc --version

Compile Custom Java Modules

Java modules (overrides) are compiled using the pinned version of the toolset.

Compile modules in the current project:

tla modules build

Compile modules in a specific project path:

tla modules build path/to/project

Verbose output:

tla modules build --verbose

The build command:

1. Resolves the project root (defaults to workspace.root from config).
2. Includes lib/*.jar files from the project root in the javac classpath.
3. Compiles .java files from the project's modules/ directory into its classes/ directory.
4. Generates the necessary Java service provider configuration for TLC overrides.

Check Java Version

tla check-java

Cache Management

The CLI caches GitHub API responses for 1 hour to prevent rate limiting. To clear this cache manually:

tla fetch-cache clear

Configuration

On first run, a default config is created at:

~/.config/tla/config.yaml

Edit this file to set your workspace path and TLC options:

tla:
  urls:
    tags: https://api.github.com/repos/tlaplus/tlaplus/tags
    releases: https://api.github.com/repos/tlaplus/tlaplus/releases

workspace:
  root: .                 # Project root (relative to CWD)
  spec_dir: spec          # Directory containing .tla files
  modules_dir: modules    # Directory containing .java files
  classes_dir: classes    # Directory for compiled .class files

tlc:
  java_class: tlc2.TLC
  overrides_class: tlc2.overrides.TLCOverrides

java:
  min_version: 11
  opts:
    - "-XX:+IgnoreUnrecognizedVMOptions"
    - "-XX:+UseParallelGC"

Directory Layout

Directory	Purpose	Location
Config	config.yaml	~/.config/tla/
Toolset Versions	Version dirs & tools-pinned-version.txt file	~/.cache/tla/tools/
API Cache	github_cache.json	~/.cache/tla/
Workspace	specs + modules + classes	Set via workspace.root in config

Note on Package Name

This package is distributed on PyPI as tlaplus-cli but imports as tla. There is a separate, unrelated tla package on PyPI (a TLA+ parser). If you have both installed, they will conflict. In practice this is unlikely since they serve different purposes, but be aware of it.

Dependencies

• Java >= 11: Required for TLC.
• uv: For installing the tool.