I need to build an internal order entry and tracking system for a grocery store which requires many of the features of existing e-commerce systems, such as product catalog, customer_to_order relations/views, movement reporting, order statuses, etc. However, the first phase of the product is purely internal, so I don't need any online e-commerce features such as shipping addresses, postal rates or a payment gateway. I've also got a bunch of business specific stuff that may not apply to a lot of on-line stores: complex product/customer discount system, lots of attributes for the products, a producer order-tracking flow (customer has an order with us and we have an order with the producer), and so on.
So I'm stuck wondering if I would be better off customizing an existing product, or rolling my own with a good web framework (such as Python/web2py)? If it was a cut-and-dry online store, then the decision would be clear - but it's not. So I'm trying to find the most extensible/flexible FOSS e-commerce software for prototyping.
The main contenders I've been considering so far are: Drupal/Ubercart, Django/Satchmo and RoR/Spree. Ubercart is undergoing a complete re-write as Drupal Commerce, so that puts me off. The Spree project looks clean and I like the ideas - but I'm already writing a product/customer ETL in Jython and don't want to balance the two languages - both Python and Ruby are new to me.
I don't like Magento's enterprise/community edition model. And I've heard lots of complaints about osCommerce and it's variants.
Thanks in advance for your thoughts.
By the way, I think the gap between the feature-set I need and what I could get out-of-the-box from an existing e-commerce product is on the order of 30%.