Change source encoding to UTF-8

[mkoncek]

We have a project which bootstraps Maven. It manually calls javac and we encountered a problem with source encoding.
I would like to unify source encodings to UTF-8.

[garydgregory]

Hello @mkoncek
How can this goal be enforced such that this build fails without the change?

[sebbASF]

Could also use:

// star extensions by Joerg Schilling

[sschuberth]

Actually, strictly (legally) speaking, "Jörg" and "Joerg" are not the same name, just like "Möller" and "Moeller" are not. So better stick to the original for credits / Copyrights, if possible.

[mkoncek]

maven-compiler-plugin is declared in commons-parent which also uses the encoding field. I rebased the PR so that the build reports an error the same way as javac would in our case. I don't know why the build doesn't fail, but there is an [ERROR] in the log if encoding is set and current sources are used.

commons-parent uses ISO-8859 encoding by default.

@sebbASF That would work too, but I believe it is time we can afford such luxury as using non-ASCII-only characters in sources.

[sebbASF]

I can confirm that changing the encoding causes the compile to report an ERROR, but the build succeeds:

$ mvn clean compile -Dcommons.encoding=UTF8
...
[INFO] Compiling 399 source files with javac [debug release 8] to target/classes
[ERROR] /commons/compress/src/main/java/org/apache/commons/compress/archivers/tar/TarArchiveOutputStream.java:[300,32] unmappable character (0xF6) for encoding UTF-8
...
[INFO] BUILD SUCCESS

I've tried experimenting with -Dmaven.compiler.failOnWarning=true (and failOnError), but Maven does not fail the build.

However, adding -Dcommons.compiler.fork=true does cause the build to fail.
Possible bug in Maven?

[sebbASF]

See https://issues.apache.org/jira/browse/MCOMPILER-491

[sebbASF]

Unfortunately, when using fork=true, some informational messages are not shown, see:
https://issues.apache.org/jira/browse/MCOMPILER-537

[sschuberth]

How can this goal be enforced such that this build fails without the change?

Maybe by adding a UTF-8 encoded test asset, plus a test that reads the file in binary mode and compares to the expected bytes?