Context:
I have a multimodules maven project that looks like:
Root ---- ModuleA
ModuleB
ModuleC
ModuleD
.......
They are around 25 modules under the Root:
A few of them represent the core of the application (5 modules)
And each of the remaining modules represent the business processes implementation related to a type a customers. These modules are completely independant among each others.
When packaging or releasing the 'Root' project, the artifact generated is a single ZIP file aggregating all the JARs related to 'Root' modules.
The single ZIP file is generated according to an assembly descriptor, it represents the delivery artifact.
At deployment time on the target environment, the single ZIP is unziped under a directory where it is consumed (class loaded) by an 'engine', a java web application that provides the final services.
Constraints
The 'business constraints' from one side,
And the willing to reduce regressions between different versions on
the other side
The above constraints lead us to adopt the following release scenarios:
Either we release the Root and ALL its submodules. It means that
the resulting ZIP will aggegate all the submodules JAR with the same
version. The ZIP will contain something similar to:
[ModuleA-1.1.jar, ModuleB-1.1.jar, ModuleC-1.1.jar, ModuleD-1.1.jar,
......., ModuleX-1.1.jar].
Or we release the Root and A FEW of its submodules, the ones that we want to re update.
The resulting ZIP will aggegate all the submodules JAR : The released submodules will be aggregated with the last released versions, the unreleased submodules will be aggregated with another 'appropriate' version.
For example, if we made a such incremental release, the ZIP will contain something similar to [ModuleA-1.2.jar, ModuleB-1.1.jar, ModuleC-1.2.jar, ModuleD-1.1.1.jar,
......., ModuleX-1.1.2.jar].
These 2 scenarios were made possible by:
Either declaring the modules as MAVEN MODULES 'module' for the first
scenario
Or declaring the modules as MAVEN DEPENDENCIES 'dependency' for
the second scenario, the INCREMENTAL scenario.
Question
Both scenarios are working perfectly BUT when we are in the 2nd scenario (INCREMENTAL), the maven-release-plugin:prepare is uploading to the SCM (svn) all the modules [ModuleA, ModuleB, ModuleD, .... ModuleX], it is uploading the released and the non released ones, whereas the 'non released modules' are declared as 'dependency' in the pom and not as a 'module'.
1/ IS THERE a way to avoid uploading the 'non released' modules ? Is there a way to inject an 'exlcude directrory list' to SCM svn provider ?
2/ A MORE global question, does the approach used is a correct one ? Or is it an anti pattern usage ? In that case, what should be the alternative ?
Thank you.
To me, your approach looks like an antipattern. I recommend to only have projects in the same hierarchy that you want to release together. Projects with a different release lifecycle should live on their own - otherwise you will keep running into the issues you mentioned. If you run the release plugin from a root directory (multi-module setup), all of the content of that root directory will be tagged in SVN.
In your case, I would probably create the following hierarchies:
Core
One per customer type
Potentially one per type to bundle them (zip), depending on your structure
I would group it by the way you create the release. It might mean that you have to run the release plugin a couple of times instead of just once when you make a change e.g. in Core, but it will be a lot cleaner.
Your packaging project will then pull in all of the dependencies and package/assemble them.
If you have common configuration options, I recommend to put them into a common parent pom. This doesn't have to be your root (multi-module) pom.
Did you try to run the maven-release-plugin with -r argument + the list of all modules you want to release?
Basically, this argument allows you to specify the list of modules against which the maven command should be performed. (if you omit it: all submodules will be included, this the default behavior)
See more details about this command line here.
I never try to use it with the maven-release-plugin, and I don't know if it will work, especially regarding SCM operations.
Related
We have a directory structure like so
java
build/build.gradle (This does NOT exist yet, but we want this)
servers
server1/build.gradle
server2/build.gradle
libraries
lib1/build.gradle
lib2/build.gradle
We have 11 servers and 14 libraries with varying uses of dependencies. EACH server is a composite build ONLY depending on libraries (we don’t allow servers to depend on each other). In this way, as our mono-repo grows, opening up server1 does NOT get slower and slower as more and more gradle code is added(ie. gradle only loads server1 and all it’s libraries and none of the other libraries OR servers are loaded keeping things FAST).
Ok, so one problem we are running into is duplication now which is why we need build/build.gradle file AND we want EVERY module in our mono repo to include that somehow for a few goals(each goal may need a different solution)
GOAL 1: To have an ext { … } section containing a Map of Strings to gradle dependencies much like so
deps = [
'web-webserver': "org.webpieces:http-webserver:${webpiecesVersion}",
'web-webserver-test': "org.webpieces:http-webserver-test:${webpiecesVersion}",
'web-devrouter': "org.webpieces:http-router-dev:${webpiecesVersion}"
]
In this way, we want ALL our projects to them import dependencies like so
compile deps['web-webserver']
GOAL 2: We want to 'include' a standard list of plugins so we are versioning all gradle plugins the same across the repo. While the above configures all jars to avoid jar hell in a mono-repo, we would like to do the same with just this section
plugins {
id 'com.github.sherter.google-java-format' version '0.9'
}
Of course, it each project may also want to add a few more plugins OR even not depend on this section(in case of an emergency and trying to just get the job done).
GOAL 3: We want checkstyle configuration (or any plugin config) to be defined the SAME for all projects (eventually!!!). We would like the checkstyle gradle to live in a common area but have all libraries somehow pull it in. Again, it would be nice for it to be optional in that, I can pull the gradle section into my build.gradle OR can create a new one in case of emergencies so I don't have to fix all projects in the monorepo right away.
IDEALLY, perhaps I kind of want configuration injection where when I run server1/build.gradle, it actually runs java/build/build.grade as it’s parent somehow but with overrides (IF I declare 'extends xxx.gradle' maybe) then all libraries it uses also use java/build/build.gradle as their parent. I am not sure this is possible or feasible. I am pretty sure 'extends xxx' doesn't exist in gradle.
Are any of these GOALS possible?
thanks,
Dean
I have been working on a monorepo with the exact same requirement as you, using gradle composite builds as well. The way we have solved this problem is by using pre compiled plugins
You need to do a new gradle project with only the code you want to share. This will create a plugin, that you can just add as a composite build and apply to the other projects.
I'm a bit confused by why you don't just use a "standard" gradle top level build file and compose the others as subprojects.
This solves all 3 of your goals
If you are concerned by build speed, you can target each server individually simply by running
./gradlew :server1:build
But if you are not able to do this for some reason you can use the apply from: syntax as described here
If I have a parent Maven project that contains the Maven wrapper files, is there any reason to also including the wrappers in the submodules of this project? Does it really matter? Any possible side effects of doing this I need to look out for? I only plan on running Maven commands from the parent project, but some day I may want to break the submodules up into their own separate projects.
Edit: To be clear, I'm talking about mvnw, mvnw.cmd, and the .mvn directory
The Maven Wrapper is only needed on the root level the project. The intention is to define a the version which will be used for building the project by downloading it automatically.
If you later decide to split up the project this would result in creating separated repositories (Git) and this means you have for each of such projects an other root which mean there it would make sense to have the maven wrapper.
I was trying to build Maven pom in something similar to the following hierarchical form:
root
+-- A-POM
+-- B-POM
+-- C-POM
+---D-POM
I was hoping that this could take care of my changed module problem. That is, if C is changed, then A must be rebuilt, etc.
But I ran into the issue that it seems the packaging at root is "pom," and after that I can't have A as packaging "war" then continue to drill in to have A include B, C as its modules. It seems to me that any POM which does not have "pom" in the then it can't have child modules. Is my understanding correct? Is there a way to do what I wanted to do?
In addition, I don't seem about to chain the "changed" mechanism in Maven (must due to my lack of knowledge). I like to have Maven detect a dependent project has changed and rebuild all the affected projects.
Thanks so much!
the reactor project (the root of the multimodule project) must have pom packaging. So your nested structure is invalid since A is not of type pom and I'm pretty sure you won't get it to work this way.
Second point is that Maven is a modularized build system and uses repository mechanisms to locate pre-built artifacts instead of checking out all modules from version control and building them in a monolithic way like in the old days ;) This means that Maven cannot know what to rebuild when you change something at your module since it simple does not have all the other module there at this time.
I think this is more a CI task than that should be handled by the build system itself. I know that your can achieve such a behavior with an appropriate build/CI Server like Jenkins that supports upstream and downstream projects. This means it is able to detect dependencies between the projects and trigger other builds as soon as a dependency has been built. This comes close to the behavior you are trying to achieve.
Btw. rebuilding other projects is only required for SNAPSHOT dependencies. Jenkins with the maven plugin supports this behavior but, depending on the number of SNAPSHOT dependencies of your project, this can cause long chains of project builds on the server. Some folks are of the opinion that in general SNAPSHOT versions are hell for CI tasks since these artifacts can change over time and are not reproducible. You could think over completely omitting SNAPSHOT versions and building final versions each time. This would also obviate your requirement to rebuild other modules as soon as a module changes. There are simply no changes until you upgrade dependency versions.
In my multi-module Maven project, suppose I have two modules, car and horse. They both depend on a JAR file, transport.jar, a file not available in any online Maven repositories. As such, I need to find a way to make these modules depend on a file found somewhere in the project folder structure.
From what I understand, the default Maven solution would be to manually register the JAR file in the local repository. While this would work on a development machine, it breaks on the build server, which clears its local repository before each build.
I've been searching online on how to do this on and off for a while and found some helpful things, but nothing that completely works.
For instance, a common answer is to add a dependency to the file using <scope>system</scope>. However, not only do others claim that it's extremely bad practice to do so, it also doesn't work on the build server. (On a side note, I would also like to point out that using absolute paths to the JAR is also out of the question due to, again, it being built on several different machines.)
A more useful method I found was to define a local repository in the POM file, pointing towards the path file:${project.basedir}/lib. (Such as in this article) Unfortunately, if I place the JAR and repository definition in the car POM, I cannot successfully add a dependency to the JAR in horse. I've tried both with and without an additional reference to car in horse, as well as defining a second repository in horse, pointing to file:${project.basedir}/../car/lib. This problem would also remain if I tried to make a third module, transport-lib, specifically for wrapping the JAR dependency.
I could most likely add the JAR file to both modules and define two separate module-local repositories, but I really don't want to unless I have to due to the need to keep the two (often updated) JARs in sync etc.
So, my question is as follows: Can someone give me a confirmed-to-work method to have two modules depend on the same JAR file inside the project, given the parameters and restrictions mentioned?
Best solution is to use a repository manager like Archiva, Artifactory or Nexus and install that artifact into the repository manager. Afterwards you can use this artifact directly in your pom files without any issue.
Don't use the scope system, cause it will cause other problem after a release for other etc.
What are the best practices for software versioning and multimodules projects with Maven?
I mean, when I create a multimodules project with Maven, what is the best approach for the versioning? To use a single version for all the modules (defined in the top project)? To use a version for each module (defined in the POM of each module)? Is there another approach that I'm missing? What are the pros and cons of each approach?
In general, are the different modules released together (possibly sharing the same version number)?
Thanks
Honestly it depends on what you would like to do. Multimodule projects are created for multiple reasons, one of them being you only need to deploy what has changed instead of all modules.
Think about it this way: if you had a non-multi-module project and you only had to change one line in the services layer, you have to rebuild the entire project and deploy all of the code again...even though only your services layer will change.
With multi-module projects, you can regenerate your project and deploy only what changed...your services. This reduces risk and you're assured that only your services module changed.
You also have a multitude of benefits to using multi-module projects that I'm not listing here but there is certainly a huge benefit to NOT keeping your version numbers of your modules in sync.
When you build your project, consider deploying it to a repository that will hold all compatible jars together for builds (each build creates a new folder with the parent-most pom version number). That way, you don't need to keep documentation about which jars are compatible...they're all just deployed together with a build number.
I was looking for a solution for this exact problem myself and versions-maven-plugin was exactly what I needed. I don't like the release plugin communicating with the SCM system. The versions plugin does just what we need: it sets a new version number in all poms of the project:
mvn versions:set -DnewVersion=2.0.0
Then I can proceed with commits, tags and an official build server build...
EDIT:
The versions plugin depends on how a maven multi-module project has been organised: as a result, it often does not update all POM files in a complex multi-module project.
I've found that sed and find do the job much more reliably:
sed -i 's/1.0.0-SNAPSHOT/1.0.0-RC1/g' `find . -name 'pom.xml'`
Typically you create a multi-module project because you have deemed that the various modules are parts of a single whole. Maybe the client-piece, the controller-piece and the services-piece. Or maybe the UI with services.
In any case, it makes sense to have the version numbers for the various modules to move in lock-step. However Maven does not enforce that as a rule.
As to your question
are the different modules released together (possibly sharing the same
version number)
I would think so. That is one of the reasons for having it a multi-module project. Otherwise you could have the modules as independent projects.
Of course this is the kind of stuff that is rife with edge cases and exceptions ;-)
I had the same problem with a project I`m working on. I also decided to use separate versions and even the dependency to the parent pom only has to be updated if some of the managed dependencies change. (so mostly as #vinnybad describes it)
Two additions
exists-maven-plugin
With the usage of the "org.honton.chas.exists-maven-plugin" only the modules will be deployed to the repository that have actually changed, which is really great, because also the according docker-images will only be published if something has changed on one of the service. This avoids "polluting" the image repository with different but unchanged versions.
versioning
One main downside of the "separated versions" approach are the questions regarding versioning:
What's the current version of my project?
Which module versions work with each other? (even thought they don't directly depend on each other, one does rely on what another does, e.g. they share the database schema)
To solve that I put all module versions into the dependency management part of the parent pom, even if no other module depends on them. A "integration-test" module could solve that by depending on all of the modules - and of course testing them together.
This way I would be "forced" to update the parent pom with every change, since it's referring the released module versions. This way the parent pom would have the "leading" version and at the dependency-management block state the versions of all modules that are compatible with each other (which will be ensured by the integration test).