I've been working on a Open Atrium project for a few months now and I wanted to get some indication from others.
The project is to create a system similar to Open Atrium, where we can have groups, users, and each group can have their features.
The reason we don’t wish to use Open Atrium is primarily because of our need for high customisation and speed. In our previous development experiments, we found that too many add-ons were needed to meet the goals, such has many alter hooks and added modules to create needed functionality, etc. This decreased the performance over time, and became cumbersome to manage.
The reason we like Drupal for this project is because of the centralized nodes. This allows for maximum customizations and virtually any type of content can be created for the system.
It may sound like a crazy question, but on a really rough estimate, what amount of time would it take to build something similar to OA completely from scratch not using any existing modules and only the base Drupal installation?
For the most part I agree it would be crazy to build our own, and we probably would use Groups, Features, Strongarm, CCK, Views, etc.
UPDATE: My intention of the question is more to identify what scale of a project OA really is, the time involved, the teams, the testing, etc. Many people don't quite understand this, as I did when I started with OA. When we realized that OA needs more and more customization for our needs, and I began building a duplicate of it, I started to think what would it take to build a similar product.
My estimate would be a team of 3-5 people, taking a total of 1500-3000 hours. However, my experience with large projects like this is non existent. So I am interested to hear what the Drupal community has to comment.