From August 21 - August 24 in Arlington Virginia, 12 people collected for a sprint, nomimally on the topic of creating a new "CMS" (content management system).

Chris Shenton was nice enough to lend his home office for the occasion. In attendance were Chris Shenton, Reed O'Brien, Tres Seaver, Chris McDonough, Malthe Borch, Carlos de la Guardia, Hanno Schlichting, Robert Marianski, Nathaniel Wingfield, Kapil Thangavelu, and Alex Clark.

On the first day, Friday, we had a long discussion about the components that might make up a not-very-simple CMS. We decided that the most important top-level components were:

• A "content repository"
• A workflow system
• An indexing system
• Observability (events)

We spent a good long while talking about the plus's and minuses of various content repository implementations. We defined a content repository as "the place where content goes", separate from any retail application which uses it. We finally decided that to get everyone's buyin that content repositories would be effectively "stupid", managed by a top-level "content manager" node that would handle indexing.

Lots of whiteboarding happened on this day.

On the second day, Malthe and Robert worked on a blog application that used "repoze.filecat" as a backend. Carlos and Chris M. and several other folks helped create a ZODB implementation of a content repository and manager system that had the (very temporary) name of "plite".

On the third day of the sprint, Malthe and Robert had finished their blog application, which used files on the filesytem to represent assets. Chris and Carlos and others had made a content system using plite that was able to add files and page objects.

Things sort of trailed off at this point, as we weren't really able to create an application that used the concepts we had come up with. Any application we could create in the time allotted seemed too simple to effectively use a content repository. So we sort of worked on infrastructure and drank beer (a keg had been strategically placed for the occasion). Malthe added a "static" directive to repoze.bfg.

Things we did well:

• Discuss and agree on components.
• Had fun.

Things we didn't do so well:

• Concentrate on coming up with any demonstration of the technology components we had dreamed up.

In future sprints, I suspect I'd like someone to come armed with an implementation of some of the components we dreamed up, with some idea about how to wire them up into a sample app that would be useful to lots of folks. We sort of bit off more than we could effectively chew by not having a more focused task at hand. But we had fun all the same. This is truly a huge task, so I wasn't particularly surprised (or even disappointed) that we didn't walk away with any truly useful software.

Many thanks to all who participated! Many thanks also to Chris' significant other, Irene, who put up with geeks in her basement and even fed us.

Pictures of the event are here .