Why is Aegir only on Drupal 5, what about a Drupal 6 version?

One of the most common questions that I get asked is why we are developing Aegir on Drupal 5. People who have been involved with the project for a while already know the answer, but I thought I'd post this executive summary.

Due to it's design, the Aegir system is fully capable of installing and managing Drupal 5, Drupal 6 and Drupal 7 sites, including managing upgrades of sites between all those versions. Only the front end (namely the hosting module, hostmaster install profile and now, the eldir theme) is Drupal 5 based.

The only people directly affected by the decision to keep the front end on Drupal 5 are us, the Aegir developers. While the decision has some negative consequences, specifically regarding integration with Drupal 6 only features, it is primarily beneficial by allowing us to focus our energies better.

In the past

When Aegir 0.1 was in active development, Drupal 6 was not an eligible target due to lack of contributed modules, among other reasons. The first release was also in development for far longer than we would have liked, so we opted to keep focused on Drupal 5 to get that release out of the door.

Even during this release Drupal 6 support was important to us, and we provided experimental support for additional platforms that allowed users to manage any Drupal 6 sites they needed to.

At present

Our primary project goal for the 0.2 release, was properly supporting the ability to manage multiple Drupal releases concurrently. Once again, due to the design of Aegir, this development primarily focussed on improvements to the back end.

Because we refactored the entire back-end for this release, by making it completely Drupal version agnostic and providing full support for Drupal 5 through Drupal 7, and also contributed heavily to the Drush 2.x release, we opted to keep the front end on Drupal 5 to avoid introducing instability and to ease maintenance of the project.
That decision is documented here : http://groups.drupal.org/node/19094

Due to this decision, we are able to provide a 0.2 release within about 4 months of the 0.1 release, with functionality that we consider absolutely crucial to our users (namely being able to manage upgrades of sites).

If we had attempted a Drupal 6 port during this development cycle, it would likely have taken many more months between releases.

Having the front end based on Drupal 5 does not negatively affect our users, because this release is specifically built to allow them to run multiple concurrent releases of Drupal, and they are free to never install any Drupal 5 sites, if they choose.

In the future

Once we have a final release of the Drupal 5 version of our 0.2 release, we are going to focus on porting the front end to Drupal 6. We will not be introducing any new features in this process, and once completed we will have a completely stable Drupal 6 - 0.2 release.

By focussing on the Drupal 6 port only after releasing the Drupal 5 version, we avoid having to keep multiple branches of 3 different projects in complete synch with each other, and we will start off with a stable code base, instead of trying to hit a moving target (the drupal 5 front end), which is built on another moving target (provision), which is built on another moving target (drush)

Work on this has already started in the community, but we will only branch once we have a final release. So if you want a Drupal 6 version, the best way to work towards it is to install and test the Drupal 5 release.

Only once we have a stable Drupal 6 version of 0.2 will we start work on the 0.3 release, whose primarily goals are oriented towards the front end, namely providing a sane user interface for managing multiple servers using Aegir.

The 0.3 release will be based on the stable D6-0.2 release, which in turn was based on the stable D5-0.2 release.

In closing

This prioritization allows us to keep our development cycles much shorter than they would otherwise have been, while not standing in the way of meeting our project goals. It also allows us to release a Drupal 6 port of the front end much sooner than we would otherwise have been able to.