[MJAVADOC-726] exclude velocity

[elharo]

Alternate approach to keeping log4j out of the tree by excluding velocity-tools.

Nuked log4j from my .m2/repo and verified that the sample project does not download log4j when generating javadoc with 3.6.1-SNAPSHOT with this patch

Tests pass locally

[michael-o]

Let's wait for CI.

[elharo]

CI being flaky. I'll rerun if needed

Recording test results

[GitHub Checks] Causes for no suitable publisher found:
[GitHub Checks] Trying to resolve checks parameters from GitHub SCM...
[GitHub Checks] No GitHub app credentials found: '5f95d117-af3b-452a-9e5c-4bddb22a67f4'
[GitHub Checks] See: https://github.com/jenkinsci/github-branch-source-plugin/blob/master/docs/github-app.adoc
[GitHub Checks] Trying to resolve checks parameters from Git SCM...
[GitHub Checks] Job does not use Git SCM
[Checks API] No suitable checks publisher found.

[elharo]

same failure as before on JDK 8 Linux:

Error: The following builds failed:
Error: * MJAVADOC-427/pom.xml

Looks like it might be an IT that's checking for very specific output.

[elharo]

Problem in IT is looking for the regexp

Link to slf4j <a href=".*?".*?><code>LoggerFactory<\/code><\/a>

but it's finding

Link to slf4j <code>LoggerFactory<\/code>

That is, the link has vanished. That might be significant.

[michael-o]

same failure as before on JDK 8 Linux:

Error: The following builds failed: Error: * MJAVADOC-427/pom.xml

Looks like it might be an IT that's checking for very specific output.

Please check these...

[elharo]

same IT failed in the same way in the effort to upgrade to 1.2.17. I still don't see how this change would cause that particular failure since neither doxia nor log4j is involved in that code path. Need to check what happens with an innocuous change

[elharo]

I do note that this failure might be caused by an unrelated problem resolving the link to a remote website that is possible from my local MacBook but not from the CI server. If so, it should repro on all PRs.

[elharo]

And now the tests are all passing. It does look like this IT is simply flaky (and likely moreso on the CI server) because it relies on external network connections that sometimes fail.