[fedora-java] Strange new build failures

[Omair Majid]

Hi,

* Roland Grunberg <rgrunber at redhat.com> [2015-04-15 13:33]:
> > jexl is now failing in rawhide with an odd failure:
> > 
> > [ERROR] Failed to execute goal
> > org.apache.maven.plugins:maven-javadoc-plugin:2.10.3:aggregate
> > (default-cli) on project commons-jexl: An error has occurred in JavaDocs
> > report generation:
> > [ERROR] Exit code: 1 -
> > /builddir/build/BUILD/commons-jexl-2.1.1-src/src/main/java/org/apache/commons/jexl2/ExpressionImpl.java:35:
> > error: bad HTML entity
> > [ERROR] * Original expression stripped from leading & trailing spaces.
> > [ERROR] ^
> > [ERROR]
> > /builddir/build/BUILD/commons-jexl-2.1.1-src/src/main/java/org/apache/commons/jexl2/Interpreter.java:888:
> > warning: no @param for node
> > [ERROR] public Object visit(ASTFloatLiteral node, Object data) {
> > [ERROR] ^
> > [ERROR]
> > /builddir/build/BUILD/commons-jexl-2.1.1-src/src/main/java/org/apache/commons/jexl2/Interpreter.java:888:
> > warning: no @param for data
> > 
> > Something odd change in java land?

Yes. We removed a patch in Rawhide that was diverging from upstream. Now
Fedora's OpenJDK 8 is more like upstream OpenJDK 8 and (proprietary)
Oracle Java 8.

> My guess is that the '- removed patch6: disable-doclint-by-default.patch'
> had something to do with it, but I haven't really checked any further.
> 
> http://koji.fedoraproject.org/koji/buildinfo?buildID=628509

Yup. This patch was *disabling* doclint. Now it is enabled by default.
This makes javadoc much more strict about what is validates as valid
javadoc. If your javadoc is invalid for some reason (bad html, unescaped
entities, adding comments after @return), then javadoc will now fail.

There's more information about this, including more details about how to
turn it off, here:
http://blog.joda.org/2014/02/turning-off-doclint-in-jdk-8-javadoc.html

TLDR: You can disable doclint by passing the `-Xdoclint:none` option to
javadoc.

Another option is to fix the upstream package to have valid javadocs and
contribute the fixes back upstream. They will hit the same issues when
they try building using the (proprietary) Oracle Java 8. This will fix
it for others too.

> I've run into the same issues. I remember a while ago hearing that
> the javadoc parser would be getting much stricter but never thought
> it'd be this strict.

Yeah, it's a bit too strict, IMHO. But upstream is unconvinced at this
point about it. Fortunately, fixing the javadoc is quite simple,
generally. If anyone needs help, I will be happy to do what I can to fix
packages.

Thanks,
Omair

-- 
PGP Key: 66484681 (http://pgp.mit.edu/)
Fingerprint = F072 555B 0A17 3957 4E95  0056 F286 F14F 6648 4681