Skip to content
This repository
file 146 lines (123 sloc) 7.441 kb

FAQ

Q: How do you pronounce Leiningen?
A: It's LINE-ing-en. ['laɪnɪŋən]

Q: What's a group ID? How do snapshots work?
A: See the tutorial for background.

Q: How should I pick my version numbers?
A: Use semantic versioning.

Q: What if my project depends on jars that aren't in any repository?
A: You will need to get them in a repository. The deploy guide explains how to set up a private repository. In general it's easiest to deploy them to a static HTTP server or a private S3 bucket with the s3-wagon-private plugin. Once the repo is set up, lein deploy private-repo com.mycorp/somejar 1.0.0 somejar.jar pom.xml will push the artifacts out. If you don't have a pom, you can create a dummy project with lein new and generate a pom from that. If you are just doing exploratory coding you can deploy to file:///$HOME/.m2/repository and the jars will be available locally.

Q: I want to hack two projects in parallel, but it's annoying to switch between them.
A: Leiningen provides a feature called checkout dependencies. See the tutorial to learn more.

Q: Is it possible to exclude indirect dependencies?
A: Yes. Some libraries, such as log4j, depend on projects that are not included in public repositories and unnecessary for basic functionality. Projects listed as :dependencies may exclude any of their dependencies by using the :exclusions key. See lein help sample for details.

Q: Why doesn't deps task populate the lib directory in version 2?
A: The only reason version 1 copied the jars around in the first place was to support existing tooling that needed a cheap way to calculate a project's classpath. Now that Leiningen has a mature plugin ecosystem, this is no longer needed; jars can be referenced directly out of the ~/.m2/repository directory. If you need to see a listing of all the dependencies that will be used and their versions, use lein deps :tree. To get the classpath use lein classpath.

Q: I specified a dependency on version X but am getting version Y; what's up?
A: One of your dependencies' dependencies has declared a dependency on a hard version range, which overrides your "soft" declaration. Running lein deps :tree will identify which of your dependencies are responsible for the version range. You can add an :exclusions clause to prevent that from affecting the rest of your dependencies. See lein help sample for how exclusions work. You may also want to report a bug with the dependency that uses hard version ranges as they cause all kinds of problems and exhibit unintuitive behaviour.

Q: I have two dependencies, X and Y, which depends on Z. How is the version of Z decided?
A: The decision depends on which depth and which order the dependencies come in the :dependencies vector: The dependency at the lowest depth will be picked. If there are multiple versions of a single group/artifact at that depth, the first of those will be picked. For instance, in the dependency graph

[Z "1.0.9"]
[X "1.3.2"]
  [Z "2.0.1"]

the direct dependency ([Z "1.0.9"]) is picked, as it is closest to the root. For the dependency graph

[X "1.3.2"]
  [Z "2.0.1"]
[Y "1.0.5"]
  [Z "2.1.3"]

the dependency X comes first, and therefore [Z "2.0.1"] is picked. If we place Y before X however, [Z "2.1.3"] will be picked.

Note that this only applies to soft dependencies, and lein deps :tree will only warn if the latest version is not chosen.

Q: I'm behind an HTTP proxy; how can I fetch my dependencies?
A: Set the $http_proxy environment variable in Leiningen 2.x. You can also set $http_no_proxy for a list of hosts that should be reached directly, bypassing the proxy. This is a list of patterns separated by | and may start or end with a * for wildcard, e.g. localhost|*.mydomain.com. For Leiningen 1.x versions, see the instructions for configuring a Maven proxy using ~/.m2/settings.xml.

Q: What can be done to speed up launch?
A: The main delay involved in Leiningen comes from starting two JVMs: one for your project and one for Leiningen itself. Most people use a development cycle that involves keeping a single project REPL process running for as long as they're working on that project. Depending on your editor you may be able to do this via its Clojure integration. (See nrepl.el or foreplay, for example.) Otherwise you can use the basic lein repl.

Q: Still too slow; what else can make startup faster?
A: The wiki has a page covering ways to improve startup time.

Q: What if I care more about long-term performance than startup time?
A: Leiningen 2.1.0 onward get a speed boost by disabling optimized compilation (which only benefits long-running processes). This can negatively affect performance in the long run, or lead to inaccurate benchmarking results. If want the JVM to fully optimize, you can follow the instructions on the Wiki page covering performance.

Q: What does "Unrecognized VM option 'TieredStopAtLevel=1'" mean?
A: Old versions of the JVM do not support the directives Leiningen uses for tiered compilation which allow the JVM to boot more quickly. You can disable this behaviour with export LEIN_JVM_OPTS= or upgrade your JVM to something more recent. (newer than b25 of Java 6)

Q: What are the downsides of Tiered Compilation?
A: Tiered Compilation sacrifices long-term JIT performance for improved boot time. Most uses of Leiningen are in a context where fast boot is more important, but in cases where this isn't the case you can switch profiles (lein with-profiles production run ...) to prevent the Tiered Compilation :jvm-opts setting from being used.

Q: I'm attempting to run a project as a background process (lein run &), but the process suspends until it is in the foreground. How do I run a program in the background?
A: For long-lasting processes, use lein trampoline run & or consider to (uber)jar the program. For short-lived ones, both lein run <&- & and bash -c "lein run &" will work fine.

Q: I need to do AOT for an uberjar; can I avoid it during development?
A: A reasonable request. Leiningen supports isolating different profiles by their target directory. Simply specify :target-path "target/%s" in order to have each profile set use a different directory for generated files. Then you can put your :aot settings in the :uberjar profile, and the .class files created from the AOT process will not affect normal development use. You can specify the profile-isolated :target-path in your :user profile if you want it applied across all the projects you work on.

Something went wrong with that request. Please try again.