Although I didn’t get the opportunity to go to the annual Umbraco meetup this year, what with a big project I’m working on, I have been following the events on the #cg12 hashtag on Twitter. And boy, am I glad I did.
A few months back, Umbraco released version 5 of their brilliant Content Management System. V5 was a pretty ambitious project. It wasn’t a straighforward upgrade, but a complete re-write of the codebase. To be honest, I was very much for the new ideas - complete re-write based on MVC; the Hive concept where different datasets were stored in a key-value schema in the Umbraco database, or any other data source for that matter. I played with many of the previews, the betas, the release candidates and, in the end, started three projects using v5.1.
As we started implementing the projects, we could see the cracks showing. It was clunky, the hive was incredibly complex and the performance was, well, horrible.
And, this morning at the Codegarden keynote, they announce that v5 is no longer being supported after 5.1. They are going back to their v4 codebase and taking the good parts of V5 and implementing them in v4.
It is frustrating that they made such a big deal of V5 and, after two years of working on a project, kill it off. I understand why they have made the decision. Umbraco 4 is a great version and we have implemented many successful projects for our clients.
I made the decision to use V5 for three projects. Two are nearly finished and we’ve gone too far to turn back. For the third, I’ve started the process of rolling back to V4.7. In hindsight I was far too quick to jump ship to an untested, unproven version that had not enough time to mature and I’ve learned a valuable lesson in my career - by all means, play with the new tech, but when there’s client money involved, assess the risks of choosing something so new so soon.
I learned from Joel Spolsky’s blog years ago that it’s a cardinal sin to rewrite - and it looks like Umbraco HQ have learned it the hard way. I applaud them for trying! I’m sure it wasn’t entirely in vain. I’m sure the future releases of v4.8+ will prove that. In fact, keep reading the posts coming out of #cg12 and you’ll see what I mean.