[Debian-astro-maintainers] Bug#919831: Javadoc -link makes broken links if module name matches package name

[Markus Koschany]

Hi tony!

Am 15.02.19 um 15:42 schrieb tony mancill:
[...]
> Hi Markus,
> 
> We independently executed identical experiments, which I'm glad for,
> because I would have wanted some external verification before uploading
> any of this to unstable.  (I started building the r-build-deps of
> libplexus-languages-java using ratt, but my laptop isn't the best place
> to build 559 packages and many packages are FTBFS right now anyway due
> to the javadoc problems.)
> 
> I believe the comment in debian/README.source about updating src:maven
> is simply out of date.  I don't see any (obvious) versioned references
> to maven-javadoc-plugin in the current maven source package.

Ok, I have just removed README.source. I think it is a bit confusing.

> So I'm not sure what the next step is, other than continuing to watch
> commits to upstream maven-javadoc-plugin and see if I can figure out why
> 3.1.0~pre0-$foo doesn't work in our environment. 

I find that really strange because I thought the latest Git snapshot
would fix the bug.

> Or now that we're into the freeze, do we want to talk about actions with
> more far-reaching consequences?
> 
> Any thoughts on whether we should focus on fixing javadoc generation or
> look at other ways to mitigate the FTBFS?

Like burning all those -doc packages? :)

In my opinion we could ask Robert Scholte for advice. He is chairman of
Apache Maven and directly involved in fixing this bug. If he doesn't
know....

However I think I have found a workaround, and we all love workarounds,
don't we.

In your initial post you pointed to a related bug report. [1] That made
me think and also read the fine Maven Javadoc documentation. There is an
option called detectJavaApiLink

https://maven.apache.org/plugins/maven-javadoc-plugin/javadoc-mojo.html#detectJavaApiLink

If I add

detectJavaApiLink=false to debian/maven.properties in libparanamer-java,
the package builds from source again.

Maybe we should patch our tools and set this property to false and move
on for now? Hopefully in a few months this will just work again without
changing this option, when maven-javadoc-plugin et al. have been
catching up?

Cheers,

Markus


[1] https://github.com/oracle/opengrok/issues/2629


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 963 bytes
Desc: OpenPGP digital signature
URL: <http://alioth-lists.debian.net/pipermail/debian-astro-maintainers/attachments/20190215/97f28364/attachment-0001.sig>