I have a largish Maven multiproject build. I'm scanning the codebase with SonarQube (5.6.5). For background, I successfully integrated the various JaCoCo exec files into SonarQube by using the "jacoco:merge" goal, to produce a single exec file. The SonarQube property that alleges to allow specifying a list of JaCoCo exec files doesn't work in our version of SonarQube, but specifying a single one does work.
I'm now trying to integrate the numerous "TEST-*" files in "target/surefire-reports" in each of the subprojects. The "sonar.junit.reportsPath" property expects a single directory, so I can't specify a list of them.
That means I'm going to have to do something as part of the build to merge the entire contents of all of the "target/surefire-reports" directories into a single directory, so I can specify that directory.
I already have a pom that does nothing but merge JaCoCo data. This seems like a logical place to store the surefire reports.
How can I do this sort of thing with Maven? Would this be an application of the "maven-resources-plugin", or something else?
Ok, well, I guess I answered my own question. I was able to get this to work with the resources plugin, specifying every one of my modules as resource directories. I now have one ridiculous-looking POM that has three lists of all of the modules in my multiproject build, for three tasks that require me to list all of the modules to process. The Gradle version would be amazingly short.
Related
I really don't have much experience in developing let alone using build tools.
I was assigned a task to build dependencies locally and get the jar files.
say I have a list of deps (GAV) like this:-
1. org.jetbrains.kotlin:kotlin-stdlib:1.6.0-RC
2. com.auth0:java-jwt:3.18.2, etc
3. openapi4j:openapi-operation-validator:1.0.7, etc
So i was able to download the source code url from maven repository and source code from github programmatically, for example :-
org.jetbrains.kotlin:kotlin-stdlib:1.6.0-RC - https://github.com/JetBrains/kotlin
com.auth0:java-jwt:3.18.2 - https://github.com/auth0/java-jwt
openapi4j-openapi-operation-validator-1.0.7 https://github.com/openapi4j/openapi4j
but there are many build.gradle files in different directories, how do I know which directory should I move into before running the gradle build command.
Things I have already tried and failed:-
For deps like this openapi4j:openapi-operation-validator:1.0.7, i can directly go into the openapi-operation-validator folder in the Github repo (https://github.com/openapi4j/openapi4j ) and run the gradle build command, but not all projects are structured like that I guess?
For deps like this com.auth0:java-jwt:3.18.2, the artifactId (java-jwt
) is already present in the github path (https://github.com/auth0/java-jwt), so i can run the gradle build command on the root github repo.
From the spring guides , among all the Gradle.build files available I can check which file has:-
jar {
archiveBaseName = <artifactId>
archiveVersion = <version>
}
, then I can move to that dir and run Gradle build, but not all build.gradle files have this.
None of the above approaches are concrete, is there any other firm approach that I can use to tackle the problem?
Your approach is generally correct.
You need to find the source code in github/gitlab/wherever, read the readme file and try to build it with whatever build tool was used there.
This may or may not work.
I have a multi project Gradle build script that runs successfully on Windows 10. It reads and updates a Version.properties file that is located away from project managed directories.
All file manipulations are done using Gradle/groovy. After the Version file has been read, incremented and rewritten it is copied to a build/classes directory where it will be picked up by subsequent jar and shadowjar tasks.
Everything works as advertised if I invoke gradle as follows:
gradle build shadowjar ... etc.
However, if I invoke the clean task prior to build the file is read and incremented properly but the copy of the file fails silently.
The command used is:
gradle clean build shadowjar
My suspicion is that gradle does not wait for the clean task to finish prior to starting the build task. The file gets read and incremented but meanwhile, the multi-project clean activities have not yet finished. I have tried variations on dependencies{} blocks, doFirst{} and doLast{} to try and push the file copy back further in the build process. My main requirement is to have the Version.properties file in place prior to the jar or shadowjar task executing. I'm suspicious of trying to write into gradle's build/ directories in that it might not be possible to put anything into the build directories while gradle is performing its activities. Is there any way to ensure that the Version.properties file (or any generated file) gets copied? Or is there another location that I can use that will not be blown away by gradle at clean time yet still get picked up in the build:jar / build:shadowjar?
You are not supposed to call gradle clean 99.99% of the time, it is redundant due to gradle's incremental build feature. So as long as you correctly define your task inputs and outputs and start from ground up in each task, the problem solves itself.
Anyway in your case the wrong order could be caused by dependency between clean and other tasks, is there any?
I have found a way to write out a generated Version.properties file that will get picked up by the jar and shadowjar tasks. Use the gradle copy task and place the revised Version.properties file into a resources directory. The build activity includes the files found in resources/ in subsequent tasks (jar, shadowjar, test, etc.) My suspicion is that because clean blows away build directories gradle assumes that the activity has fully completed when it starts the build. I think that I've proven that this is not the case. doFirst{}, doLast{} and dependencies{} do not seem to work as modifiers to clean build.
Current my scanner is running through and only scanning the parent and skipping the rest of my nested files. If I run sonarlint (using the cli and specifying some test and source files) , it tries to analyze 37k files instead of the few I need. I have been able to skip ~3k files by adding the <sonar.skip>true</sonar.skip> property to a pom file. However, I still can't configure the project to run across certain sub-folders and print out some kind of code coverage test. (Is JaCoCo needed for the latest version(6.3,0)? Or can code-coverage be handled through some configuration?).
If Sonar seems to be analyzing too many files, it is probably because you had not set the sonar.sources=src/main/java in your sonar-project.properties file, so it defaults to the basedir and includes everything.
SonarQube can't do code-coverage itself, it just reports on coverage-reports from a tool like JaCoCo. It is funny they don't clarify these things in https://docs.sonarqube.org/display/SCAN/Analyzing+with+SonarQube+Scanner --but with enough digging, you can find good info on that site.
I have a main war file with maven dependencies to various war files. With the maven-war-plugin I am able to merge them into one, but my problem is, if files with the same name exist they will be overridden by the last one.
What I want is to merge the files with the same name.
This is not something done commonly with Maven. But have a look at the discussion of this SO question: How can I merge resource files in a Maven assembly?
In the gradle documentation says:
Builds which utilize a buildSrc directory will generate a second
profile report for buildSrc in the buildSrc/build directory.
How can we do that (utilize build/Src) via the gradle sript, couldn't you help me?
You may put your helper scripts/classes to various places. One of them is buildSrc directory.
See below quote from gradle documentation.
When you run Gradle, it checks for the existence of a directory called
buildSrc. Gradle then automatically compiles and tests this code and
puts it in the classpath of your build script. You don't need to
provide any further instruction. This can be a good place to add your
custom tasks and plugins.
Your qoute only tells that if you use buildSrc directory, you will have second profile report.