Comments on: Iteration 0

By: jcteague

jcteague — Tue, 15 Jul 2008 17:51:28 +0000

@Ryan,
I personally do not put my deployments in my CI, but I’ve been on projects that do and it is nice. But they certainly ar automated.

Right now, my project releases to QA every week (at the end of each iteration) and i do that with a simple bat file to run the necessary nant targets.

The reason why I don’t put them in my CI is that I now have a deployment for each of my revisions, and on a big project that could be several a day. I feel that it is too granular.

I worked with the guys on the Tarrantino project, and they built a deployer that is based upon the svn revision. And I hold the same opinion about that approach. SVN revision numbers are too granular. I prefer tagging my QA deployments and having a limited set of rollback points. Nightly builds I could see, but haven’t automated yet.

Thanks for the feedback

By: Ryan Cromwell

Ryan Cromwell — Tue, 15 Jul 2008 17:31:41 +0000

To expand on Jason’s comment and your CI task, I find it to super helpful to include Deployment in your CI. Nightly builds are a good first step, but do it as part of your CI builds and you’ll have flawless upgrades/installs in the future.

I personally also like to give my clients access to run the CI builds. There is no better way to close the feedback loop.

By: Jason Meridth

Jason Meridth — Tue, 24 Jun 2008 15:02:17 +0000

Your 3 are my usual. Infrastructure is the other one. I ensure that the servers, PCs and network is in place. If it can’t be done I make it a release task that has to monitored closely.

We just got done introducing 3 new technologies into our iteration A this release. Some teams introduce new technologies in Iteration 0 by refactoring current items to use the new technology. One example, is to put dependency injection via Windsor or StructureMap in place. It still follows the definition of refactoring because the client/end user doesn’t notice a difference.

Excellent post John.