lein-monolith
is a Leiningen plugin to work with multiple projects inside a
monorepo. At a high level, the plugin gives you a way to:
- Share configuration across subprojects, such as
:repositories
,:managed-dependencies
,:env
, etc. - Run tasks across a multiple projects matching sophisticated selection criteria.
- Run tasks across a globally-merged view of multiple projects.
- Query dependencies, generate graphs, and other utilities.
For a more detailed introduction to the project and some motivation, see this 2016 Seajure presentation.
Library releases are published on Clojars. To use the latest version with Leiningen, add the following plugin to your user profile or project definitions:
The monolith
task provides several commands to make working with monorepos
easier. In order to use them, you'll need to create some configuration telling
the plugin where your subprojects are.
The configuration is provided by a metaproject, which lives in the repository
root and must contain a value for the :monolith
project key. Create a
top-level project.clj
file and add the plugin and
monolith entries.
See the configuration docs for more details about the available options.
lein-monolith
can be used inside the individual projects within the monorepo,
or you can use it from the repository root to operate on all the subprojects
together.
Many tasks which operate on multiple projects accept targeting options, which generally select or filter the command to a subset of the subprojects.
:in <names>
Add the named projects directly to the targets.:upstream
Add the transitive dependencies of the current project to the targets.:upstream-of <names>
Add the transitive dependencies of the named projects to the targets.:downstream
Add the transitive consumers of the current project to the targets.:downstream-of <names>
Add the transitive consumers of the named projects to the targets.:select <key>
Use a selector from the config to filter target projects.:skip <names>
Exclude one or more projects from the target set.
Each <names>
argument can contain multiple comma-separated project names, and
all the targeting options may be provided multiple times.
To see a list of all the projects that lein-monolith knows about, you can use
the info
task:
lein monolith info [:bare] [targets]
This will print out the config file location, coordinates of every subproject
found, and a relative path to their location within the repo. For scripting, you
can pass the :bare
flag, which will restrict the output to just the project
name and path.
The plugin also provides the deps-on
task to query which subprojects have a
certain dependency:
lein monolith deps-on lib-b
Or you can go the other way with deps-of
to find the subprojects which a
certain project depends on:
lein monolith deps-of app-a
A useful higher-order task is each
, which will run the following commands on
every subproject in the repo, in dependency order. That means that projects
which don't depend on any other internal projects will run first, letting you do
things like:
lein monolith each check
lein monolith each :upstream :parallel 4 install
lein monolith each :start my-lib/foo do check, test
lein monolith each :select :deployable uberjar
In addition to targeting options, each
accepts:
:start <name>
provide a starting point for the subproject iteration.:parallel
run subproject tasks concurrently, up to the number of specified threads.:endure
continue applying the task to every subproject, even if one fails. If any projects fail, the command will still exit with a failure status. This is useful in situations such as CI tests, where you don't want a failure to halt iteration.:report
show a detailed timing report after the tasks finish executing.:silent
suppress task output for successful projects.:output
path to a directory to save individual build output in.
The :refresh
option only visits projects that have changed since the last
:refresh
. This allows incrementally building your projects:
lein monolith each :refresh ci/build install
The project is only considered refreshed if the task is successful. This means you can run tests over the projects that have changed since the last successful test run:
lein monolith each :refresh ci/test test
Behind the scenes, lein-monolith is storing hashed fingerprints of each project, which you can inspect and manually manipulate:
lein monolith changed
lein monolith mark-fresh :upstream ci/build
lein monolith clear-fingerprints ci/test
The plugin can create a profile with :resource-paths
, :source-paths
and
:test-paths
updated to include the source and test files from multiple
projects in the monorepo. The profile also sets :dependencies
on each internal
project, giving you a closure of all dependencies across all the subprojects.
This can be useful for running lint and tests on multiple subprojects at once:
lein monolith with-all [targeting] test
The link
task creates
checkout
symlinks to all the internal packages that this project depends on. The plugin
also includes an unlink
task as a convenience method for removing checkout
dependencies.
To link checkouts for all transitive dependencies, you can pass the :deep
option.
If you have existing checkout links which conflict, you'll get warnings. To
override them, you can pass the :force
option.
lein monolith link [:deep :force] [project...]
lein monolith unlink [:all] [project...]
In general, it's recommended to only link between the projects you're actually actively working on, otherwise Leiningen has to recursively trace the full tree of checkouts before running things.
Licensed under the Apache License, Version 2.0. See the LICENSE file for rights and restrictions.