Skip to content
Documentation/canton-network-docs/CLI ToolsView on canton-network-docs
canton-network-docs/CLI Tools

dpm (Daml Package Manager)

dpm is the command-line interface for Canton development. It manages SDK installation, project scaffolding, compilation, testing, code generation, and local development environments. Most Canton development workflows start with a dpm command.

Pre-requisites

Dpm runs on Windows, macOS and Linux. For full functionality, you must have installed:
  1. VS Code download
  2. JDK 17 or greater, installed and part of your JAVA_HOME. If you do not already have a JDK installed, try OpenJDK or Eclipse Adoptium.
  3. If you are on a Mac with Apple Silicon, you may need to install Rosetta 2 to run x86 binaries. You can do this by running the following command in a terminal:
softwareupdate --install-rosetta

Installation

Installing dpm usually involves downloading a single binary and adding it to your PATH. When installing dpm (any edition, on any platform), you can set the DPM_HOME environment variable to change the location where the SDK and any future updates are installed. The default is:
  • ${HOME}/.dpm/ on Mac and Linux
  • %APPDATA%/.dpm/ on Windows
You should then add ${HOME}/.dpm/bin to your PATH:
export PATH="${HOME}/.dpm/bin:${PATH}"
or, if you used a custom install location:
export PATH="${DPM_HOME}/bin:${PATH}"
To install the latest version:

Mac/Linux Installation

curl https://get.digitalasset.com/install/install.sh | sh

Windows Installation

Download and run the windows installer, which will install the dpm sdk and set up the PATH variable for you.

Manual Installation

If you prefer a more manual installation process, see Manual Installation Instructions.

Installing dpm without an SDK

You can obtain the bare dpm binary — which does not have any SDK bundled with it — in two ways. Download it from the releases page. Or pull it directly from the public OCI repository using oras (or a similar tool):
## use the oras CLI to pull the correct platform image:
oras pull --platform darwin/arm64 europe-docker.pkg.dev/da-images/public/components/dpm:latest

## make the binary executable:
chmod +x dpm
The example above pulls the macOS Apple-silicon image. Substitute the --platform value for your machine: linux/amd64, linux/arm64, darwin/amd64, darwin/arm64, or windows/amd64. Once you have the bare binary, you can run dpm install as described in Project Management: with no arguments it is an alias for dpm install package, installing the current project’s dependencies; with a version argument (dpm install <version>) it installs that SDK version. Because the bare binary is not bundled with an SDK, you must set the registry manually — see Registry and Edition.

Upgrading the SDK

You can upgrade to the latest version of an SDK by running:
dpm install latest
This installs the latest SDK of the same edition you already have installed. You will not be required to provide credentials again, since you should already be configured from installing the SDK the first time. You can also install a specific version of the SDK in a similar way:
dpm install <version>
Note that outside of a Daml project, the active SDK is the latest version that is installed.

Verifying the Installation

Once installed, verify the installation:
dpm --version

Configuration Files

daml.yaml

Every Daml project has a daml.yaml file at its root. If a daml.yaml file doesn’t exist in your project, you can create one with:
dpm init
It specifies the SDK version, project metadata, source directory, and dependencies.
sdk-version: 3.4.0
name: my-project
source: daml
version: 1.0.0
dependencies:
  - daml-prim
  - daml-stdlib
build-options:
  - --target=3.4
Key fields:
  • sdk-version — The Daml SDK version to use. dpm install downloads this version.
  • name — The package name, used in the output DAR filename
  • source — Directory containing Daml source files (typically daml)
  • dependencies — List of Daml library dependencies
  • build-options — Compiler flags (e.g., target LF version)

multi-package.yaml

For projects with multiple Daml packages, a multi-package.yaml at the project root lists all sub-packages:
packages:
  - ./contracts
  - ./tests
  - ./scripts
Running dpm build --all from the root builds all listed packages in dependency order.

dpm-config.yaml

The dpm-config.yaml file stores global dpm configuration such as SDK installation paths and registry settings. It is stored at ${DPM_HOME}/dpm-config.yaml (where ${DPM_HOME} is usually ~/.dpm), is optional, and rarely needs manual editing. dpm can be configured through this config file and through environment variables simultaneously. Environment variables take precedence over the corresponding fields in dpm-config.yaml, so they can be used to override configured values. Typical uses include:
  • Registry URL — Override the default OCI registry for SDK components.
  • Authentication — Point to registry credentials.
  • Insecure registry — Allow HTTP connections to a registry.

Registry and Edition

For many of its built-in commands, dpm needs to know which OCI registry is in effect. If dpm was obtained as part of an SDK, the registry is set automatically in ${DPM_HOME}/dpm-config.yaml (where ${DPM_HOME} is usually ~/.dpm). However, if you are using the bare binary obtained directly from its OCI repository, this configuration might not be known. You can set it manually in ${DPM_HOME}/dpm-config.yaml:
registry: europe-docker.pkg.dev/da-images/public # or your desired OCI registry
or using the DPM_REGISTRY environment variable.

Authentication

Authentication for pulling from and pushing to OCI registries is based on a file similar to ~/.docker/config.json, though a Docker installation on the system isn’t necessary. dpm defaults to using Docker’s auth. You can configure Docker to authenticate to the OCI registry (Google Artifact Registry), and dpm will use that:
gcloud auth login
gcloud auth configure-docker europe-docker.pkg.dev

Variable interpolation

Both daml.yaml and dpm-config.yaml support variable interpolation, which lets you avoid hardcoded values (such as registry URLs or credentials paths) and reference environment variables or other dynamic values.

Command Reference

Project Management

dpm init — Initialize a new Daml project in the current directory. Creates a daml.yaml configuration file and a basic project structure. You can manage SDK versions manually by using dpm install. To install the SDK version specified in the daml.yaml, run:
dpm install
To install a specific SDK version, for example version 3.5.2, run:
dpm install 3.5.2
To see the active SDK version:
dpm version --active
3.5.2
To list the installed SDK versions, including the currently active one (marked with *):
dpm version
  3.5.1
* 3.5.2
To additionally list all the SDK versions that can be installed, as well as the installed versions:
dpm version --all
To get the list in a machine readable format:
dpm version --all -o json
 [
    {
        "version": "3.4.11",
        "installed": true,
        "remote": true
    },
    {
        "version": "3.5.1-rc3",
        "remote": true
    },
    {
        "version": "3.5.1-rc5",
        "remote": true
    },
    {
        "version": "3.5.1",
        "installed": true,
        "remote": true,
        "active": true
    },
    {
        "version": "3.5.2",
        "remote": true
    }
]
dpm init
dpm new — Create a new project from a template. Templates provide working examples that you can modify.
dpm new my-project --template daml-intro-contracts

Building and Testing

dpm build — Compile Daml source files into a DAR (Daml Archive) file. The output goes to .daml/dist/.
dpm build
For multi-package projects, build all sub-packages:
dpm build --all
dpm test — Run all Daml Script tests in the current package. Produces a summary with pass/fail status and coverage report.
dpm test
dpm install — Download and install the Daml SDK version specified in daml.yaml.
dpm install

Code Generation

dpm codegen-js — Generate TypeScript/JavaScript bindings from a compiled DAR file.
dpm codegen-js .daml/dist/my-project.dar -o generated-ts
dpm codegen-java — Generate Java bindings from a compiled DAR file.
dpm codegen-java .daml/dist/my-project.dar -o generated-java

Development Environment

dpm sandbox — Start a local Canton sandbox node. This runs a single-participant Canton instance with an in-memory ledger for testing.
dpm sandbox
dpm studio — Open Daml Studio (the VS Code extension) for the current project. This launches VS Code with the Daml extension activated.
dpm studio

Exploration

Use --help on any command to see available options:
dpm --help
dpm build --help
dpm codegen-java --help
dpm inspect-dar --help

Summary of dpm Commands

  • dpm build: Build a Daml package or project This builds the Daml project according to the project config file daml.yaml (see configuration files). In particular, it will use the dpm SDK (specified in the sdk-version field in daml.yaml) to resolve dependencies and compile the Daml project. Given a daml.yaml and .daml source files, the dpm build command will generate a .dar for this package. See How to build Daml Archives for how to define a package and build it to a DAR.
  • dpm test: Test the current Daml project or the given files by running all test declarations. This runs all daml scripts defined within a package. Daml Scripts are top level values of type Script (), from the daml-script package. This package mimics a Canton Ledger Client for quick iterative testing, and direct support within Daml Studio. The command runs these scripts against a reference Ledger called the IDE Ledger, which implements the core functionality of the Canton Ledger without the complexity of multi-participant setups. It is most useful for verifying the fundamentals of your ledger model, before moving onto integration testing via the Ledger API directly, or the Daml Codegen. dpm test also provides code coverage information for templates and choices used.
  • dpm clean: Clean a Daml package or project This removes any Daml artifact files created in your package during a daml build, including DARs.
CommandDescription
dpm addAdd components and dars to project
dpm add componentAdd a component to project
dpm add darAdd or update a dar in the project
dpm component initInitialize a component in the current directory
dpm component runPull down and run a remote component
dpm updateUpdate project dependencies
dpm codegen-javaDaml to Java compiler
dpm codegen-jsDaml to JavaScript compiler
dpm canton-consoleCanton console client
dpm daml-shellDaml-shell client for PQS
dpm damlcCompiler and IDE backend for the Daml programming language
dpm docsGenerate documentation for a daml package from its documentation comments
dpm initInitialize a daml.yaml project configuration file in the current directory
dpm installInstall project’s dependencies or specific dpm-sdk version
dpm install packageInstall the SDK(s), components and dars used by current project
dpm uninstallUninstall a dpm-sdk version
dpm versionShow installed sdk versions
dpm inspect-darInspect a DAR archive
dpm newCreate a new Daml package
dpm publishCommands for publishing artifacts
dpm publish componentPublish a component to an OCI registry
dpm publish darPublish a dar to an OCI registry
dpm pqsParticipant query store
dpm sandboxRun full Canton installation in a single process
dpm scriptDaml Script Binary
dpm studioLaunch Daml Studio
dpm tagsList published tags of an artifact
dpm upgrade-checkCheck upgrade validity between package versions
dpm validate-darValidate a DAR archive
Note: after installing a specific SDK version with dpm install <version>, update your project config file to use the new version.

Components

Components are artifacts that dpm (being primarily a launcher) wraps. Examples of components are sandbox and scribe. A component defines the commands it wants dpm to expose, as well as the binaries that back those commands. A component is a directory that contains:
  • a component.yaml file at its root. Its main purpose is to tell dpm how to import the component’s commands and exec them.
  • the binary or binaries (or JARs) that back the commands defined in component.yaml.
  • optionally, additional arbitrary files and directories (if needed for the operation of the component, for example lib directories). When the component is published, dpm includes these in the artifact (see Publishing Components to an OCI Repository).
Components — or rather the commands in them — can be backed by native binaries and/or JARs. This is a sample component.yaml defining two commands (foo and bar) backed by the same native binary:
spec:
  commands:
    - path: ./foo-binary
      name: foo
      aliases: ["f"]
      desc: a command for foo'ing!
    - path: ./foo-binary
      name: bar
      aliases: ["b"]
      desc: a command for bar'ing!
      exec-args: ["--should-bar-instead-of-foo"]
When running dpm --help, you’ll see the following in the help message:
Usage:
  dpm [command]
...
...
Dpm-SDK Commands
  foo    a command for foo'ing!
  bar    a command for bar'ing!
...
The exec-args on the bar command tells dpm to supply additional args when running the binary backing the command, so it is executed as ./foo-binary --should-bar-instead-of-foo.

Dependency on other components

You can optionally declare dependencies of your component on other components:
# component.yaml

dependency-paths:
  foo: CUSTOM_FOO_PATH
spec:
  commands:
    ...
In this example, the component expects a component named foo to be present when running dpm. dpm injects an environment variable named CUSTOM_FOO_PATH into this component’s environment, pointing to the path of foo on the filesystem. The dependency components must be present either as part of the installed SDK, or declared in daml.yaml or multi-package.yaml.

component.yaml schema

See the JSON schema.

Working with Components

Initializing a component

Use dpm component init to scaffold a new component in the current directory:
dpm component init
This creates a starter component.yaml (and a daml.yaml) in the current directory. Pass --force to overwrite existing component.yaml and daml.yaml files:
dpm component init --force

Running a remote component

Once a component has been published, you can pull it down and test-run it with dpm component run:
dpm component run <component name> <component version> [command] [args]
The registry to pull from is the one configured via the DPM_REGISTRY environment variable or ${DPM_HOME}/dpm-config.yaml. This is handy for quickly sanity-checking a published component:
export DPM_REGISTRY=europe-docker.pkg.dev/da-images/public
dpm component run canton-open-source 3.3.0-snapshot.20250429.15795.0.v6b2dcccb
dpm pulls the component into ~/.dpm/cache/components/, and then runs the specified command of your component.

Adding components and DARs to a project

Use dpm add component to add a published component to your project:
dpm add component <oci-uri>
OptionDescription
--insecureUse HTTP instead of HTTPS for the OCI registry
Use dpm add dar to add or update a DAR in your project. You must specify whether the DAR is a dependency or a data-dependency:
dpm add dar <oci-uri> <--dependencies | --data-dependencies>
OptionDescription
--dependenciesAdd the DAR to the dependencies field
--data-dependenciesAdd the DAR to the data-dependencies field
--insecureUse HTTP instead of HTTPS for the OCI registry

Publishing

Publishing Components to an OCI Repository

This functionality is available in DPM version 1.0.14 or later (or bundled with SDK 3.5 or later).
dpm fully handles packaging and publishing components. Components are published as OCI artifacts to an OCI registry. As a component developer, you can use the dpm publish component command to publish a component. Publishing is platform-aware, and the command requires the target platform(s) to be specified with -p/--platform:
--platform darwin/arm64=<path to prepared directory containing the component for this platform>
Multiple platforms can be published in a single command by providing multiple --platform flags. For JAR components, you can use --platform generic=<path to dir>. Here’s a sample component foo laid out with one directory per platform:
./
  darwin-arm64/
        component.yaml
        foo              # darwin/arm64 binary
        foo2             # another binary
        lib/...          # directory
        somefile.txt
  linux-amd64/
        component.yaml
        foo
        lib/...
The component.yaml file is required, and must be present at the root of each platform directory. Publish it like this:
dpm publish component \
  oci://<full destination URI>:<version> \
  --platform darwin/arm64=./darwin-arm64 \
  --platform linux/amd64=./linux-amd64 \
  --extra-tags latest    # also push a 'latest' tag for this version of the component
This runs some validations, then publishes two OCI images corresponding to the two platforms, and an OCI index, plus the latest tag. The published artifact for a given platform includes everything present in the given directory for that platform — all files and subdirectories (for example lib). You don’t have to pre-bundle or zip up anything; just provide the raw directory. When pulling a component, dpm restores the artifact exactly as it was at publish time. The component can now be imported (pulled) and run by dpm. See Running a remote component for how to pull components.

Options

OptionDescription
-p, --platformREQUIRED. <os>/<arch>=<path-to-component> or generic=<path-to-component>. May be given multiple times.
-t, --extra-tagsPublish extra tags besides the semver.
-a, --annotationsAnnotations to include in the published OCI artifact (key/value pairs, may be given multiple times).
-g, --include-git-infoInclude git info as annotations on the published manifest.
--authPath to a config file similar to Docker’s config.json for authenticating to the OCI registry. Defaults to Docker’s config.json.
-d, --dry-runDon’t actually push to the registry.
--insecureUse HTTP instead of HTTPS for the OCI registry.
When publishing build artifacts you can also apply annotations to the entire image. The -a/--annotations flag can be used with a key/value pair multiple times to add annotations to the image being published. This is useful for attaching metadata such as the branch the build artifacts were built from, or the commit SHA related to the artifacts.

Publishing DARs to an OCI Repository

This functionality is available in DPM version 1.0.20 or later (or bundled with SDK 3.5.2 or later).
Similarly to components, you can publish DARs to an OCI repository for distribution:
dpm publish dar oci://example.com/my/dars/foo:1.0.0 \
    -f ./foo.dar \
    --extra-tags devnet    # optional
You should now be able to see the published version via dpm tags; see Listing available tags and versions below. See the Remote Dars docs section to learn how to install oci:// DARs in your project.

Options

OptionDescription
-f, --darREQUIRED. Path to the DAR file to publish. May be given multiple times.
-l, --licensePath to the LICENSE file. A license file is required when publishing a DAR.
--exclude-licenseFor non-production use only. Disable the license file requirement for DAR publishing.
-t, --extra-tagsPublish extra tags besides the semver.
-a, --annotationsAnnotations to include in the published OCI artifact (key/value pairs, may be given multiple times).
-g, --include-git-infoInclude git info as annotations on the published manifest.
--authPath to a config file similar to Docker’s config.json for authenticating to the OCI registry. Defaults to Docker’s config.json.
-d, --dry-runDon’t actually push to the registry.
--insecureUse HTTP instead of HTTPS for the OCI registry.

Listing available tags and versions in an OCI repository

You can use the dpm tags command to view published versions and tags for a specific component or DAR:
dpm tags oci://example.com/my/dars/my-package
1.0.0
1.2.3
latest
OptionDescription
--registry-authPath to a config file similar to Docker’s config.json for authenticating to the OCI registry. Defaults to Docker’s config.json.

Typical Workflow

A common development cycle with dpm:
# 1. Set up the project
dpm new my-app --template daml-intro-contracts
cd my-app
dpm install

# 2. Write Daml code, then compile
dpm build

# 3. Run tests
dpm test

# 4. Start a local sandbox for integration testing
dpm sandbox

# 5. Generate bindings for your backend
dpm codegen-java .daml/dist/my-app.dar -o generated-java
dpm codegen-js .daml/dist/my-app.dar -o generated-ts

daml assistant to dpm migration steps

This section provides a step-by-step guide for projects originally built with the Daml assistant.

Migration steps

  1. Install Dpm Follow the installation instructions above.
  2. Verify Dpm is in your PATH
dpm version
  1. Remove ~/.daml/bin from your PATH Edit your shell configuration file (e.g., ~/.bashrc, ~/.zshrc, ~/.profile) and remove any line that adds ~/.daml/bin to your PATH
  2. Remove the Daml assistant
rm -rf ~/.daml
  1. Create a project configuration if needed If a daml.yaml file doesn’t already exist in your project, generate one:
dpm init
  1. Replace daml commands with dpm Most commands map directly. See the command migration table below. Six daml commands have been removed and replaced with Declarative API, Canton Console, JSON API, and/or gRPC API calls. See removed command replacements for full details.
  2. Review project configuration Review your daml.yaml to ensure it is compatible with dpm. Audit parent directories for stale daml.yaml files. Delete orphan or invalid files.
  3. Review global configuration (optional) Review or create ${DPM_HOME}/dpm-config.yaml if you need to customize registry, authentication, or other global settings.
  4. Set up variable interpolation (optional) Use variable interpolation in your configuration files to avoid hardcoded values.
  5. Update CI/CD pipelines (if applicable) Update any CI/CD pipeline scripts that reference daml commands to use the corresponding dpm commands instead.

Command migration table

daml commanddpm commandpurpose
daml newdpm newCreate a new Daml project
daml builddpm buildCompile the Daml project
daml testdpm testRun tests for the Daml project
daml installdpm installInstall Daml SDK components
daml codegen javadpm codegen-javaJava code generation
daml codegen jsdpm codegen-jsTypeScript code generation
daml damlcdpm damlcInvoke the daml compiler
daml studiodpm studioOpen project in Visual Studio
daml sandboxdpm sandboxLaunch a Daml Sandbox
daml ledger allocate-partiesUse Declarative API – OR – JSON / gRPC APIAllocate parties on a ledger
daml ledger list-partiesJSON / gRPC APIlist parties on a ledger
daml ledger upload-darUse Declarative API – OR – JSON / gRPC APIUpload (and vet) dars on a ledger
daml ledger fetch-dargRPC APIFetch a Dar from a ledger.
daml packagesJSON / gRPC APIPackage a Daml project
daml startdpm sandbox + dpm build; use Declarative API – OR – JSON / gRPC to upload/allocateStart a local Daml Ledger

Removed command replacements

The following daml commands have no direct dpm equivalent. Use the Declarative API, Canton Console, JSON API, or gRPC API instead.

daml ledger allocate-parties

MethodDetails
Declarative APIcanton.parameters.enable-alpha-state-via-config = yes with canton.participants.sandbox.alpha-dynamic { parties = [{ party = "Alice", synchronizers = ["mysync"] }, { party = "Bob" }] }
Canton Consoleledger_api.parties.allocate(...)
JSON APIPOST /v2/parties/
gRPCPartyManagementService.AllocateParty

daml ledger list-parties

MethodDetails
Canton Consoleledger_api.parties.list()
JSON APIGET /v2/parties or GET /v2/parties/{party}
gRPCPartyManagementService.ListKnownParties

daml ledger upload-dar

MethodDetails
Declarative APIcanton.parameters.enable-alpha-state-via-config = yes with canton.participants.sandbox.alpha-dynamic.dars = [{ location = "./my-asset.dar" }, { location = "https://path.to/repo/token.dar", request-headers = { AuthenticationToken : "mytoken" } }]
Canton Consoleledger_api.packages.upload_dar(...)
JSON APIPOST /v2/dars/
gRPCPackageManagementService.UploadDarFile

daml ledger fetch-dar

MethodDetails
JSON APIGET /v2/packages/{package-id}
gRPCPackageService.GetPackage

daml packages

MethodDetails
Canton Consoleledger_api.packages.list()
JSON APIGET /v2/packages or GET /v2/packages/{package-id}/status
gRPCPackageService.ListPackages

daml start

Replace with dpm sandbox combined with dpm build. Use the Declarative API or JSON/gRPC API to upload DARs and allocate parties as needed. See dpm sandbox for details on running a local Canton installation.
  • Daml SDK — What the SDK includes and how versions work
  • Sandbox — Local testing environment started by dpm sandbox
  • Daml Studio — VS Code extension launched by dpm studio
  • Daml Script — Writing and running tests with dpm test

Manual Installation Instructions

Mac/Linux Installation

If you cannot / wish not to use the shell script to install for Linux or OSX, you can alternatively install dpm manually by running this set of commands in your terminal: The latest stable release version can be found by hitting the following URL:
VERSION="$(curl -sS "https://get.digitalasset.com/install/latest")"
And you can then use this to retrieve the tarball of the full installation, extract, and install, as outlined in the full instructions below.
#! /usr/bin/env bash
set -xeuo pipefail

# get latest version number
VERSION="$(curl -sS "https://get.digitalasset.com/install/latest")"

# set your architecture to either amd64 | arm64
ARCH="$(uname -m | sed 's/x86_64/amd64/;s/aarch64/arm64/')"

# set your OS to either darwin or linux
OS="$(uname | tr '[:upper:]' '[:lower:]')"

# pull down appropriate tarball for your OS and architecture
readonly TARBALL="dpm-${VERSION}-${OS}-${ARCH}.tar.gz"

# determine location of tarball to download
TARBALL_URL="https://get.digitalasset.com/install/dpm-sdk/${TARBALL}"

# make tmpdir
TMPDIR="$(mktemp -d)"

# download tarball
curl -SLf "${TARBALL_URL}" --output "${TMPDIR}/${TARBALL}" --progress-bar "$@"

# create directory to extract into
extracted="${TMPDIR}/extracted"
mkdir -p "${extracted}"

# untar to extracted directory
tar xzf "${TMPDIR}/${TARBALL}" -C "${extracted}" --strip-components 1

# bootstrap dpm
"${extracted}/bin/dpm" bootstrap "${extracted}"

# verify dpm version runs and does not return an error
if ! "${extracted}/bin/dpm" version; then
  echo "dpm version failed" >&2
  exit 1
fi

# cleanup tmpdir
rm -rf "${TMPDIR}"

Windows Installation

Download and unpack the windows dpm-sdk archive (.zip), then:
# Extract the downloaded zip ($ZIP_PATH) to temp directory ($EXTRACTED)
# Avoid using the system's temp directory as the user may not have rights to it
New-Item -ItemType Directory -Path $EXTRACTED | Out-Null
Expand-Archive -Path $ZIP_PATH -DestinationPath $EXTRACTED

# Optionally, override the TMP and DPM_HOME environment variable to point to directories other than the default,
# as the user might not have rights to the default directories.
# (You might also want to persist these variables as DPM uses them on every invocation)
$env:TMP = "<user-owned temporary directory>"
$env:DPM_HOME = "<user-owned directory>"

& "$EXTRACTED\windows-amd64\bin\dpm.exe" bootstrap $EXTRACTED\windows-amd64

Unstable Version Manual Installation

Preview / unstable versions are also available for experimentation, though it is always recommended to use the stable versions listed above instead.

Unstable Mac / Linux

Follow the dpm-manual-installation-mac-linux instructions to install an unstable version, but note the differences in VERSION and TARBALL_URL. The latest unstable release version can be found by hitting the following URL:
VERSION="$(curl -sS "https://get.digitalasset.com/unstable/install/latest")"
The unstable tarball can be retrieved from the following URL:
# determine location of tarball to download
TARBALL_URL="https://get.digitalasset.com/unstable/install/dpm-sdk/${TARBALL}"

Unstable Windows

Download and unpack the latest windows unstable dpm-sdk archive (.zip), then follow dpm-manual-installation-windows instructions.