Phly, boy, phly
the weblog and site of Matthew Weier O'Phinney

Zend Framework 2.0 Roadmap

2.0 marks the first new major release of Zend Framework, and, as such, is the first time we can break backwards compatibility. Major releases are generally of two flavors: large, new featuresets, or rewrites and refactoring to fix architectural concerns. ZF2 falls primarily in this latter category.

I posted a rough roadmap in November on the ZF wiki, to which we received a lot of feedback. Several ideas we brought up that were contested, and we re-considered many of the decisions and goals we outlined as a result before we started implementation.

Also, around 5 or 6 weeks ago, I started a discussion with Bill Karwin, who led the project from mid-2006 through the 1.0 release in 2007 and slightly beyond. He had some solid feedback on the nature of the roadmap, and with this information, results of a poll we did late last year, and feedback I've had via mailing lists, IRC, twitter, blogs, and more, I published a new roadmap that focussed less on implementation detail while firmly and succinctly stating the requirements for the project.

Stated in a sentence:

The primary thrust of ZF 2.0 is to make a more consistent, well-documented product, improving developer productivity and runtime performance.

The basic goals are as follows:

• Ease the learning curve
• Make extending the framework trivially simple
• Improve baseline performance of the framework
• Simplify maintenance of the framework
• Be an exemplar of PHP 5.3 usage
• Provide mechanisms for using just the parts of the framework needed

We also stated several general development objectives for those contributing to ZF2:

• Simplify
• Programming by Contract
• Favoring the Explicit

For more detail on each of these goals, I encourage you to read the document.

What has been accomplished

While the roadmap has only really stabilized recently, that does not mean we haven't been working steadily on its development. There were some objectives we anticipated as early as 2 years ago. Among these were migrating the project to namespaces, providing infrastructure to allow cherry-picking components for packaging, and updating the unit test infrastructure to make better use of more recent PHPUnit features.

Shortly after 1.10.0 was released, I created a temporary git repository on my own server, and started work. The first task I did was to update the unit test suite and analyze all class files for dependencies to assist in the namespaces migration.

After completing this process, my entire team -- all three of us -- started the work of migrating the code to namespaces. Ralph wrote a tool that scanned the library and created a map file of existing classes and suggested namespace/classname combinations. We then used this tool as a launching point for the migration, each of us working on a component at a time. This work was by no means automated -- we discovered very quickly that such a tool only took care of the most cursory work. I detailed some of our findings a couple months back; we ran into a number of issues we never anticipated, and the progress has been far from speedy. At this point, however, we have migrated everything but the Zend_Service classes, the MVC, and those components that build on top of the MVC (Application, Navigation, Form, etc.).

We also rewrote a few components during this time, as we discovered inconsistencies or in areas where we had problems with unit testing. One such is a component that has been a pain point basically since its creation: Zend\Session. The new design gives a good idea of what can be accomplished during a focussed rewrite, and by using 5.3 features where they make sense, and I'm very pleased with how it turned out.

In parallel with this effort, I also did a fair bit of research determining how we would offer our Git repository and workflow. We're going for a fairly traditional workflow where only a small handful of developers will have commit access, and all other contributors will submit pull requests to those developers -- for everything ranging from documentation fixes to bugfixes and feature topics. To ensure that those contributing have signed a CLA, we have created a pre-receive hook that verifies either the author or reviewer against a list of CLA signees. Additionally, we have created post-receive processes to create RSS feeds and deliver email notifications. These processes will be easy for us to hook into to add new functionality -- such as sending updates to a twitter account, performing subtree merges, and more. This should aid us greatly in setting up continous integration in the near future.

The official Zend Framework 2.0 Git repository is available for cloning:

• git clone git://git.zendframework.com/zf.git

The helpful folks at Github have also kindly provided a mirror of the repository; our hope is that contributors can fork from there in order to collaborate on new features and bug fixes. (I've also cloned it under my own Github account, for those who want to issue pull requests.)

Warning! Zend Framework 2.0 development is in very early stages, and should not be used for developing production applications. In fact, the APIs will change in the coming weeks and months, and should not be relied on for really much of anything.

Community Initiatives

A number of new initiatives have sprung up in the last week surrounding the community's involvement in the ZF2 process.

The primary goal of my team has been to get the library migrated sufficiently so that we can open the repository to cloning, and allow contributors to begin working on initiatives to improve the framework. This is now possible, and several other initiatives are emerging.

Several community members have put forth the idea of a community review team. This effort is still taking shape, but the basic goals are:

• Assist contributors in getting patches into the framework, primarily by acting as a liaison to missing maintainers or arbitrators between maintainers and other contributors.
• Shepherd new feature proposals into the master branch, by performing proposal review and code review.

For more information on this effort, please review the thread on the zf-contributors mailing list.

A number of contributors are also starting to discuss rewrites and refactoring of components. Much of this is being done on the zf-contributors mailing list, and some on the #zftalk.dev channel on Freenode. If you are interested in contributing, I highly recommend subscribing to the list and dropping into the channel when you can.

End Notes

These are exciting times for Zend Framework development -- the first time we can break backwards compatibility since the 1.0 release, and a chance to participate in cutting edge PHP 5.3 development. While the process has been slow, it's also been incredibly rewarding and a huge learning experience -- and I'm glad I've had the chance to participate. I hope you'll join us!