Maven is great for breaking gigantic code bases into many little modules, with
identified dependencies. This allows incremental builds etc.
And then, to run that code, we usually put all together again into a
uber-mega-JAR or WAR, or with mile-long class paths. While sometimes this makes
sense, often it does not.
diet4j can run command-line apps, and Tomcat web apps, similarly to how maven
builds projects: simply specify the name of the top project, and diet4j
assembles all the other JARs automatically for the run.
E.g. if your project hierarchy looks like this:
Project A
Project B
Project C
Project D
you can say:
> diet4j A
which will read the POM in A.jar (in ~/.m2/repository, or a location of your
choosing), determine run-time dependencies, then recursively look for B.jar,
C.jar and D.jar, load them into separate ClassLoaders, hook up dependencies and
jump on the main program (if it has one) of Project A.
No jar-with-dependencies and stuff like that required. Usually no changes are
required to maven projects or program structure.
As I said, it may not be for all people, but it is for some who want to
distribute changes incrementally, avoid rebuilding gigantic JARs every time
some small change needs to happen, fit better into Linux-style package
management etc. It’s also great for dynamically finding and loading modules
without restarting the application.
I’d love some feedback, it’s early days.
http://diet4j.org/ <http://diet4j.org/>
Thank you,
Johannes Ernst
