Blog

Archive for November, 2008

Merb 1.0 Spec Change

By | November 23rd, 2008 at 5:11PM

When we released 1.0, we also copied the existing 1.0 specs into a directory marked spec10 and added a rake task called spec:oneoh. This allows us to make sure that new versions of Merb still run against the same API as Merb 1.0, and that we’re not accidentally breaking working 1.0 APIs (to the extent that our 1.0 specs can make such an assertion).

The rule is that those specs must not be changed, with one exception. Specifically, we’re allowed to modify the 1.0 specs if the spec itself broke for reasons unrelated to the thing it was attempting to spec, and the breakage does not indicate a breakage in the 1.0 API. Additionally, any such change requires a public announcement, to minimize the number of such changes.

As a result, I’m announcing a small change in spec10, and the rationale for that change.

Many people have accidentally overridden core Merb::Controller methods in their controllers, with unpredictable (and confusing) results. From my Rails days, I remember this happening as well. As a result, Merb, starting with 1.0.2 and 1.1, will raise an error if you attempt to override a Merb::Controller method.

One of our tests had a controller which used the “method” action. Of course, “method” is already a method on all objects, so our new code caused it to raise an error. However, the test was not asserting that it was legal to override “method”; it was the same sort of accident that caused us to add this feature in the first place. As a result, we have gone into the spec10 directory and modified the action name to “request_method” instead of “method”, allowing 1.0.2 to pass the 1.0 specs.

And that requires this public post.

I hope that these posts will be few and far between, but at the very least, they show a commitment to the public API from the Merb core team that I hope you find reassuring. Thanks!

Popularity: 1% |

Scotland on Rails

By | November 18th, 2008 at 5:11PM

We’re proud to announce that we’ve just signed on as a platinum sponsor for the upcoming Scotland on Rails. This year’s event takes place in Edinburgh on March 26 – 28, and includes a tutorial charity event with Rails luminaries Chad Fowler and Marcel Molina.

We had the pleasure of working with Paul, Alan and Graeme last year, and really enjoyed our interactions with their team. The community feedback after the event was overwhelmingly positive, and things look like they’ll be even better this year.

The Call for Papers is open: give it a shot!

Popularity: 1% |

Next: MerbDay Atlanta

By | November 17th, 2008 at 5:11PM

For those of you who couldn’t make it to MerbCamp, or who are looking for another dose, MerbDay Atlanta‘s the place to go.

On Saturday, December 6th, members of the Merb Core Team will be gathering at the Georgia Tech Hotel and Conference Center in Atlanta for an event slightly more intimate than MerbCamp. The crowd will be smaller, and several of the sessions will be in-depth tutorials, designed to help folks get up and running.

Of course, you’ll still hear from the bigwigs: Engine Yard’s Yehuda Katz will be delivering the closing keynote. It’ll be a great opportunity to get a peek at the future of Merb, and the changes and developments to come.

Time’s running out: Register Now!

Popularity: 1% |

Rubinius: Past, Present and Future

By | November 17th, 2008 at 5:11PM

As part of the continuous process of guiding Engine Yard’s future, we must make certain that we’re allocating resources properly in the near term so that our longer term projects can continue on to fruition.

A few weeks ago at RubyConf 2008, Koichi Sasada, the lead developer of Ruby 1.9 (also known as YARV) did Rubinius the honor of projecting that Rubinius would eventually be the Ruby interpreter of choice. We agree — however, we also agree that it will take a while to get there.

So today we made a substantial change in the Rubinius team sponsored by Engine Yard and reduced the team to two members from its previous six. This was a *hard* decision to make, but it is clear that it’s a good choice. We *must* fund Rubinius in a rational way that allows us to continue funding it to completion.

By funding Rubinius for almost two years now, including *heavy* funding in 2008, we helped it get from a good idea to something substantial and real.

The RubySpec suite is now a separate project and is being used by Gemstone, Sun, and Microsoft in their own Ruby implementations. Tens of thousands of software executable tests were developed during this time that specify what Ruby is, exactly: something that had never been expressed before outside of MRI itself and Dave Thomas’ outstanding book. This is very good, because it means that Ruby, the language, will not differ substantially from platform to platform.

Rubinius moved toward Ruby completeness to the point where it ran Rails, the RubyGems system, and other significant software projects such as Merb at RailsConf 2008. Since that major milestone they rewrote the VM in C++, refactoring and testing along the way, so that Rubinius is now on a firm architectural base.

Finally, we’re delighted to be working with Gemstone to refactor the Rubinius kernel (the part written in Ruby), to allow using it as a shared component between Rubinius and their MagLev runtime.

Engine Yard remains 100% behind Rubinius, and congratulate the team for their phenomenal results to date.

Popularity: 1% |

Merb 1.0 Released

By | November 8th, 2008 at 5:11PM

Merb 1.0 Final was announced at RubyConf yesterday. This is a major milestone for Merb, and a big step for the community. Congratulations to all the Merb contributors who put so much time and effort into this project!

You can get the new install on RubyForge at http://rubyforge.org/projects/merb/, or directly from the Merb project gem server:

gem install merb-source http://edge.merbivore.com

Popularity: 1% |