Hudson and Jenkins: Two Weeks Later

Most Java developers have probably heard about the recent Hudson/Jenkins split. InfoQ’s most recent article on the topic, by Alex Blewitt, ends on an upbeat note:

With the commercial support of both Oracle and Sonatype behind the development of Hudson, the future looks good for the eponymous continuous integration tool. However, Jenkins continues to evolve as well [etc, etc]…

I’m not so sure about Hudson’s future. Fortunately, there are a number of easily measurable indicators that should allow us to gauge the progress of these two rival projects.

First, we can look at commit counts. Specifically, commit counts should provide a decent leading indicator of the health of each project, given that they generally foretell whether or not the project is likely to address user needs in new releases.

Second, we can look at user mailing list post counts, which generally tell the story of user uptake after all of the hard work has gone into fixing bugs, adding features and cutting new releases. I would consider user mailing list usage to be a lagging indicator.

So after two weeks, what does the commit picture look like? Well, Hudson has seen 40 commits since the split [1], while Jenkins has seen 166 commits. It’s not even close. As a leading indicator, this bodes well for Jenkins’ future.

What about user mailing list activity? Remember that logically this should be a lagging indicator. As such, I expected the Hudson user list to remain more active than the Jenkins user list for quite some time, even if Jenkins ends up overshadowing Hudson over time. Astonishingly, the Hudson user list has seen 55 posts since the split, while the Jenkins user list has seen 563 posts in the same time frame! Have users already decided, en masse, to adopt Jenkins over Hudson? That’s certainly the picture painted by these numbers.

Two weeks is an extremely small track record on which to base any conclusions, and some of these numbers may be distorted by infrastructure issues on the Hudson side, but I think it’s safe to say that Hudson’s future is a little murkier than Alex’s optimism might suggest. The good thing is that we’ll be able to revisit these numbers in 6 months and have a pretty clear idea of where things are headed :-)

[1] All of the counts in this post are as of the time of writing, and begin counting on February 1st, 2011.

UPDATE (4/4/2011): I’ve posted a newer commit graph here.

About these ads

12 Comments

  1. Hank said,

    February 16, 2011 at 6:18 am

    While the actual degree of Jenkins’ dominance may be a surprise, I can’t say that the dominance itself is. From the beginning, I’ve thought Oracle made a mistake with their handling of the situation and that the VAST majority of Hudson users would switch to Jenkins as soon as the fork was official.

  2. Alex Blewitt said,

    February 18, 2011 at 10:43 am

    Thanks for the link. I agree with you that the pace of development is markedly different between the two. Just imagine what it would look like if Sonatype hadn’t stepped in.

    @alblue

  3. Werner Keil said,

    February 18, 2011 at 10:49 am

    Something similar as with JavaFX Scripting, now a Google Code Project, abandoned by Oracle (to its luck maybe, so there is no split or fork there, the community will decide whether they like it or not, Steven Chin gets his chance tomorrow in Chennai ;-)

  4. Brian said,

    February 18, 2011 at 3:44 pm

    I wonder how many of those jenkins mail posts were about how to make the switch. They will sharply drop off soon

  5. Menomuna said,

    February 18, 2011 at 4:31 pm

    Also, check out kinds of email discussions going on projects. Discussion on Hudson list is interesting — even with “only” 55 messages, discussion is mostly on things only tangentially related to development or usage.

  6. February 18, 2011 at 10:04 pm

    55 posts and how many of those were either ‘unsubscribe’ requests or plugin developers declaring intention to drop support for Hudson.

    Even the Marquess of Queensberry would have called this one by now.

  7. February 19, 2011 at 3:51 am

    And the IRC activity? Yesterday I joined #hudson by acident. Old haabit. And were kicked of immidately becaus it’s an invite only channel O_o

  8. Illiad said,

    February 21, 2011 at 1:04 am

    Hudson dev mailingl ist is mostly about high-fives between Oracle and Sonatype guys.
    If you Winston, Duncan, and Susan are part of Oracles. While if you google the other names + Sonatype, usually you’ll find relationship there.

    It’s ironic that Sonatype is driving the development of Hudson after Ted Farell got very sensitive about his team’s technical ability. The Oracle guys ended up as admins.

    The only thread that seems to contain differing point of views between Oracle and Sonatype, http://java.net/projects/hudson/lists/dev/archive/2011-02/message/214, doesn’t have a resolution. I’m guessing it’s something they ended up discussing privately.

    You call that a success?

  9. February 21, 2011 at 7:51 am

    […] I’m sure I had to know about this already, but Jenkins: The Definitive Guide is an open-source book on everyone’s favourite politically embroiled CI server. Speaking of politics, here is an interesting graph. […]

  10. February 21, 2011 at 1:51 pm

    In all fairness, a large number of the commits to Jenkins were likely changes in the code and documentation changing “Hudson” to “Jenkins”.

    That being said, I have switched from Hudson to Jenkins, and don’t expect to switch back.

  11. Joni said,

    February 21, 2011 at 6:48 pm

    Craig: actually, the (new) Hudson codebase *also* contains “Jenkins” -> “Hudson” commits, as, technically, that seems to be the way the fork went. See https://twitter.com/vbehar/status/37561522821537792

  12. Mick said,

    March 10, 2011 at 12:16 pm

    Nice metrics…keep them coming !!!


Follow

Get every new post delivered to your Inbox.

%d bloggers like this: