Jira Git Plugin

Appreciate the attention. Let me know if I can help in any way.

I suspect what you need is to make your settings.xml to include the following repositories (see end of comment) - alternatively a profile.xml could be added to project, which I can't do this minute. Settings.xml should be in your ~/.m2 (or M2_REPO or whatever) and the rest of the file looks like http://confluence.atlassian.com/display/DEVNET/Example+settings.xml

I validated this by doing mv ~/.m2 ~/.m2backup, I copied the settings.xml file form .m2backup to .m2, then

mv ~/.m2 ~/.m2backup # so I don't use files you don't have
cp ~/.m2backup/settings.xml ~/.m2 # see below and link
svn co http://svn.atlassian.com/svn/public/contrib/jira/jira_git_plugin/
cd trunk 
mvn -Pplugin-debug

Now I'm missing:

• com.jcraft:jsch:jar:0.1.37 # version has changed repository has moved?
• javax.transaction:jta:jar:1.0.1b # should be in sun's - a case issue ? 1.0.1b vs. 1.0.1B?
• javax.activation:activation:jar:1.0.2 # instructions are included

Javax jars are typically painful, you generally need to find them on sun and inject them into the maven repository as detailed when you run mvn. Any of these if you find the jar then mvn tells you how to inject it into the maven repository, so its not so bad for just 3 files.

                <profile>
                        <id>jnolen-devnet-profile</id>
                        <activation>
                                <activeByDefault>true</activeByDefault>
                        </activation>
                        <repositories>
                                <repository>
                                        <id>atlassian-public</id>
                                        <url>https://maven.atlassian.com/repository/public</url>
                                        <snapshots>
                                                <enabled>true</enabled>
                                        </snapshots>
                                        <releases>
                                                <enabled>true</enabled>
                                        </releases>
                                </repository>   
                                <repository>
                                        <id>atlassian-m1-repository</id>
                                        <url>https://maven.atlassian.com/maven1</url>
                                        <layout>legacy</layout>
                                </repository>
                                <repository>
                                        <id>sun java maven repository 2</id>
                                        <url>http://download.java.net/maven/2</url>
                                </repository>
                                <repository>
                                        <id>iBiblio Maven2</id>
                                        <url>http://mirrors.ibiblio.org/pub/mirrors/maven2/</url>
                                </repository>
                        </repositories>
                        <pluginRepositories>
                                <pluginRepository>
                                        <id>atlassian-public</id>
                                        <url>https://maven.atlassian.com/repository/public</url>
                                        <snapshots>
                                                <enabled>true</enabled>
                                        </snapshots>
                                        <releases>
                                                <enabled>true</enabled>
                                        </releases>
                                </pluginRepository>
                        </pluginRepositories>
                        <properties>
                                <downloadSources>true</downloadSources>
                                <downloadJavadocs>true</downloadJavadocs>
                                <!-- 
                                  Uncomment this and specify the location on your file system where your 
                                  clover license is located.
                                  <clover.licenseLocation>/path/to/clover/license</clover.licenseLocation>
                                -->
                                <atlassian.pdk.server.url>http://localhost:1990/confluence/</atlassian.pdk.server.url>
                                <atlassian.pdk.server.username>admin</atlassian.pdk.server.username>
                                <atlassian.pdk.server.password>admin</atlassian.pdk.server.password>
                        </properties>
                </profile>

I don't have time to check this now, but once those are injected, re-running mvn -Pplugin-debug should start a webserver on localhost:1990/confluence and you can start to play with the developer environment.

You're welcome.

I see what you mean in regards to the bare repository thing. I somehow missed that there was a code block right underneath which was supposed to find this. This does however not work if you give a path like the following as root: /git/Repository.git/, which I would consider being a correctly named bare repository.
Maybe it was better if the repository root path was stripped for any trailing slashes, and then checked if the path ends on .git. That way we would know if the user is trying to address a bare repository or not, and could improve the logic based on that. What is your thoughts on that?

By the way, another thing I would like to have made, but I could figure out how to, was to have the repository name parsed in the URLs for webgit, as ${rev} does. Then you could simply set up a "template" for webgit links, and use that for all Git repositories you may have - as long as they are located on the same server. I need to read up on Velocity in order to find out how to expose the repository name to the template.

I had a go at making the findGitDir a bit more intelligent, in order to make it find the correct directory in the first guess. Here's the code for the function:

	private static File findGitDir(String root) {
		File current = new File(root).getAbsoluteFile();
		if (current.getName().endsWith(".git")) {
			if (log.isDebugEnabled())
				log.debug("checking for gitdir: " + current.getAbsolutePath());
			if (isGitDirectory(current) != null)
				return current;
		} else {
			File gitDir = new File(current, ".git");
			if (log.isDebugEnabled())
				log.debug("checking for gitdir: " + gitDir.getAbsolutePath());
			if (isGitDirectory(gitDir) != null)
				return gitDir;
		}
		return null;
	}

So it simply checks if the path given ends with .git, and if it does then it checks if the directory pointed to is a Git folder, otherwise it checks the path + .git as before. I don't really see a need for the function to try to find the .git directory on it's own, since people 'should' be able to specify the correct path in my opinion

Regarding your note about not having to specify the repository origin when adding the repository, then you could probably use the org.spearce.jgit.lib.Config functions or so in order to find the origin. Note though in regards to this, than when you have a bare repository there is no origin path, so it should probably not be specified for bare repositories.

I hope my (limited) work can help make this plugin better.

The git hook approach seems to address the problem JIRA Git Plugin has with tracking multiple branches.  I do agree the large amount of noise generated in the JIRA comments would be a problem.

I'm very new to JIRA, so this question might be phrased incorrectly, but would it be possible to use only the seperate tab feature from the JIRA Git Plugin, and then use a Git hook to actually populate the tab?

"it appears like you are using sarcasm"

Not really. Who knew placement of quotes can imply so much?

Kidding aside, thanks for your comments.

I think whether you use a Jira plugin or a git-hook to integrate Jira with git depends on your usage model. In our case, we were quite used to using Trac with SVN and were very happy with it. Once we switched to Jira and git, I tried to continue emulating the Trac model, and hence the scripts.

I don't really see "commit noise" as noise. Obviously its a matter of personal preference and/or company policy, and some people will disagree. In our case, we really would like to get notified every time a commit is pushed to our "public" git repository and git-jira-hook solves this problem as well. When a git push is made, the series of commits updates the Jira ticket(s), which in turn causes Jira to send out a nicely formatted email. By clicking on links embedded in that email, the recipient can now look at the exact git commit(s) as well.

To answer your other question "You do comment for commits contained in the merge right?" -- I don't think I understand your question clearly. In the event the "merge" caused a commit, this commit would have to have the Jira magic markers embedded or else the commit would not succeed. Does that answer your question, or were you asking something else.

I've ported the changes to 0.5-snapshot - I'm getting a null pointer on a repo id during retrieval of commits for an issue. As I mentioned above, I'd seen that you removed the repoId from the lucene document, so maybe this line should be commented out or something.

(In my tests I'm running a single repository with many branches and 2 projects. The git repository is actually our live repo, but the issues are all fabricated and imported from a csv file and are pretty bogus. The master branch case worked, but the error was an commit/issue NOT on master and was also the other project.)

I'll try to get this going this weekend - I'm willing to sacrifice multi-repo in the short term, and I'd be willing to put this out as it is as I think it immediately adds value.

The features that (I think - maybe I'm missing something) need to be added are the ability to scope the branches searched (it takes over 5 minutes on my repo easily for a scan, but it is scanning EVERY branch afaict) and showing which branch(es) the commit is in in the git commit tab. (I'd still be willing to release this without those features however).

I did sacrifice the repository id for the branches