I was able to run clirr on uimaj-core by using the 2.2.3-SNAPSHOT version
(adding
<pluginRepositories>
<pluginRepository>
<id>codehaus-plugins</id>
<url>http://snapshots.repository.codehaus.org</url>
</pluginRepository>
</pluginRepositories>
to your .m2 settings.xml file to enable maven to find it, if you want to try).
The 2.2.2 version gives up with some kind of internal error. 2.2.3 still gets
the error, but "continues".
It produced a bunch of duplicate, intermingled lines of output, which, after
sorting and then removing the duplicates, gave the following and changes in
"public APIs":
(o.a.u == org.apache.uima]
[ERROR] o.a.u.analysis_engine.impl.PrimitiveAnalysisEngine_impl: Method
'protected o.a.u.analysis_engine.ResultSpecification
computeAnalysisComponentResultSpec(o.a.u.analysis_engine.ResultSpecification,
o.a.u.resource.metadata.Capability[])' has been removed
[ERROR] o.a.u.cas.impl.AllowPreexistingFS: Method 'public
o.a.u.cas.impl.AllowPreexistingFS[] values()' is now final
[ERROR] o.a.u.cas.impl.TypeSystemUtils$PathValid: Method 'public
o.a.u.cas.impl.TypeSystemUtils$PathValid[] values()' is now final
[ERROR] o.a.u.cas.TypeClass: Method 'public o.a.u.cas.TypeClass[] values()' is
now final
The first: the method to compute the result spec was removed from the
...impl.PrimitiveAnalysisEngine_impl class to the
ResultSpecification_impl class. These are both impl classes, so this is not in
the public API.
The previous method is marked "protected" - but I know of no instances of
someone subclassing ResultSpecification_impl.
The 2nd, 3rd, and 4th all seem related to enum impl, so may be bogus.
The 2nd: This is a new enum - it is public because it is referenced from
multiple packages.
But it is in an ".impl" package, so is not part of the public API.
The 3rd: This is in the public API - in the sense that there are other projects
which refer to it (SimpleServer, DictionaryAnnotator...).
Nothing changed, though.
The 4th: This is referenced in two different packages, so must be public.
Again, though, nothing changed.
So all is ok.
It would probably be good to run clirr on other projects, too... and wade
through the results...
However, I'll probably defer that work until some point in the future...
-Marshall
