Satchmo is turning out to be an interesting project. The first true Python contender to challenge the plethora of PHP CMS tools used for the de facto net store front, it fares quite well in the face of this challenge. .
Satchmo is based on the Django framework. More accurately, it is loosely coupled to Django through several external Django plugins. For example, registration functionality is extended through the django_registration external interface, installed as a separate egg in the Python distribution directory. Satchmo makes calls to this extension, leaving Django code untouched.
Compare this to Drupal registration changes, where source code is wedged into the existing framework, applying deep back end database schema changes, as well as code changes throughout the framework. The changes are irreversible, and if they fail, you’re SOL. Look under the hood of Drupal, and you see a tangle of database calls, auth and group checking calls, intertwined around back end logic. It’s not pretty, which is why it’s so hard to maintain and upgrade over time.
The Satchmo/Django coupling is not perfect. If you mismatch incompatible versions of Django plugins to Satchmo, or other applications, cryptic messages can appear from either side, pointing back to code compiled elsewhere and placed in your Python distro. This makes it difficult to trace back, even if you vaguely know where the failure point is.
But this is the worst problem Satchmo has presented to me so far. I can certainly live with that, compared to digging through many chunks of Drupal framework source code and database schema backups to figure out why a recent patch broke authentication, how group permissions were changed, etc.
Satchmo takes full advantage of all of the juicy Django goodness that makes web framework development fun again: built-in internationalization and localization all the way down to currency handling and language choices in templates, built-in registration, email verification and customer account management, built-in form data validation, seamless form-to-database data entry.
This is a revolutionary, much needed improvement in the Open Source store front choices. Now all we need is an Open Source back end inventory system, and a very functional Open Source phone bank system based on Asterisk, with Python wrappers, and life would be wonderful.