Created attachment 126622 [details]
Showing the error and the patch in action

Created attachment 126623 [details]
Patch for git

Created attachment 126624 [details]
Patch for hg

Created attachment 126625 [details]
Patch for svn

I like to contribute patches to solve this issue. The modules git/mercurial/subversion are affected. (there had to be 3 patches because of the open task http://netbeans.org/bugzilla/show_bug.cgi?id=213679)

Cause of error: 2 links (author/issue linker) were added with the same id and the variable of author was reused. 

@NetBeans-Dev: Please review, discuss with me and commit!

Hi, thanks for noticing this, the evaluation and the patches, they were helpful. However the problem is slightly somewhere else. It seems you do not have bugtracking.bridge module installed (or enabled). Without this module (where we implement the abstract class VCSHyperlinkProvider) the following block of code will be skipped:
> for (VCSHyperlinkProvider hp : providers) {
>     l = IssueLinker.create(hp, hyperlinkStyle, master.getRepositoryRoot(), doc, commitMessage);
>     if (l != null) {
>         linkerSupport.add(l, 0);
>         break;
>     }
> }
and "l" will be still set to an instance of AuthorLinker.class as you correctly realized.

> Cause of error: 2 links (author/issue linker) were added with the same id and
> the variable of author was reused.
It's no problem in this case. As far as the instances are of different classes it won't matter, take a look into VCSHyperlinkSupport.add - there's a list of linker instances that we associate with a specific ID. In case of these tooltips the ID can be still the same. In fact it should be the same IMHO. You see, the different ids make sense e.g. in case of a search history view, where you have a list of revisions, each revision/row has a specific id and we associate a set of linker instances (one author, one messaage etc.) with each revision=ID.
And if you take the tooltip window you actually have just one revision - so the id should be the same. I know this pseudo API is difficult to comprehend, it is sometimes difficult even for the authors.

The correct fix here is trivial, simply add "l = null;" before the above mentioned code and it will work (i hope). Please make this modification, try and test it and attach the patches again. This time please merge the three modifications into one single file, i will commit and push your changes as one commit anyway.

Created attachment 126638 [details]
Patch - resetting l to null (git-hg-subversion)

(In reply to comment #6)
> The correct fix here is trivial, simply add "l = null;" before the above
> mentioned code and it will work (i hope). Please make this modification, try
> and test it and attach the patches again. This time please merge the three
> modifications into one single file, i will commit and push your changes as one
> commit anyway.

Here the updated trivial patch. It works for me, tested with hg.

IMHO reusing a local variable is ugly. So i will provide another patch which does not reuse the local variable. You may decide which one you want to commit.

Created attachment 126639 [details]
Patch - no reuse of local variable - use braces (git-hg-subversion)

Patch which uses braces instead of reusing a local variable

thanks for the patch, i just pushed it into the repository: http://hg.netbeans.org/core-main/rev/1735ebb8228f

Integrated into 'main-golden', will be available in build *201210300001* on http://bits.netbeans.org/dev/nightly/ (upload may still be in progress)
Changeset: http://hg.netbeans.org/main-golden/rev/1735ebb8228f
User: markiewb@netbeans.org
Log: Issue #220923 - Blame tooltip of GIT/Mercurial/Subversion does not show commit message when team server plugin is enabled
stop reusing the author linker if no bugtracking system is activated