annotation-processors | Java Annotation Processor sample code | Build Tool library

So I managed to fix the problem with the maven-shade-plugin. I added following piece of code to my pom.xml file:

Try this:

1. Do not set a PDF Font Name for the font family that you created with the Arial ttf. You don't want to use Helvetica because it doesn't include glyphs for Vietnamese characters. Leave it empty so that the Arial ttf is used in PDF.
2. Set the PDF Encoding for the font family to Identity-H (Vietnamese uses horizontal writing).
3. The screenshot shows that the font family is called "times", so you'll have to use for the text element.

This will result in the Arial ttf being embedded in the PDF file (a subset of it actually) and used for the text. Therefore the PDF would show the Vietnamese characters, as long as the Arial ttf includes glyphs for them (and my Arial ttf copy does).

If for some reason you don't want to embed Arial in the PDF file, you can also use the DejaVu fonts included in net.sf.jasperreports:jasperreports-fonts:6.9.0, for instance

You might also want to set the net.sf.jasperreports.default.pdf.encoding property to Identity-H, if you generally want to embed fonts in PDFs, or just leave it unset and specify the PDF encoding for each font family.

Since latest Gradle version ( >= 4.8 I would say) you can simplify your build script as follows ; you don't need apt plugin anymore, just use annotationProcessor Gradle configuration :

One way to solve this issue is use some sort of a registry in between builds, you can for example store the types annotated in a service like style in meta-inf

So in your processor you defer the code generation until the processing last round, and after generating your code you store the types in a file into a file under the META-INF

And as I know there is no way to control the order of processors execution.

I did not look into how either Dagger 2 or AssistedInject work under the hood, but in general Annotation Processing is done in multiple rounds.

Annotation processing happens in a sequence of rounds. On each round, a processor may be asked to process a subset of the annotations found on the source and class files produced by a prior round.

Files get generated during those rounds, but validation (and errors) happens on a later or last round, when every file was generated and processed. This is also mentioned on a second answer on the question you linked. Someone also wrote a detailed answer about how rounds work on a different question.

Apparently I just had to wipe pitest from my maven repo. It repulled the jars and it worked. It seemed to have been holding onto settings, even when I had removed PITest entirely from my poms.

edit: apparently I'm an idiot. I had mvn clean org.pitest:pitest-maven:mutationCoverage which of course wipes my target directory before attempting to find the classes.

IntelliJ IDEA (for this question Community 2018.2) will sometimes generate an invalid configuration for a JUnit5 test.

To see the run configuration from the menu choose: Run -> Edit Configurations...

In this example there are two run configurations for Class CalculatorTests generated by IntelliJ IDEA.

The first configuration generates the exception:

Instead of using Intellij IDEA, you should configure your build process entirely in gradle. This way it will be IDE-independent, and IDEA supports auto-sync with the gradle project.

In gradle you can try something like this, and then run the gradle 'build' task (or 'classes' task to only compile source):

I resolved this by removing the -proc:none compiler argument from the compiler plugin. With that present none of the generated sources were being compiled at all. With the 3.1 plugin I had to have that, but with 3.6.0 I can't.

I also tried to implement the answer https://stackoverflow.com/a/35045416/4756238 by making the compilerArg specific to the default-compile phase, but that didn't compile the generated sources. If I didn't reuse the default-compile id the build worked and gave me generated class files, but it ran two compile phases, with the -proc:none one second, which seemed redundant.

The final pom section for the compiler looks like this: