We’ve been dependent on Django 1.6 for our components, for many reasons. There are a lot of things Review Board has to deal with that most Django apps do not, so Django often regresses us, unintentionally. There are also just design changes in Django that don’t suit shipping products, and we’re often having to work around these changes.
However, the biggest bit is our database migration story. Our codebase depends on Django Evolution for migrations, which has to dive pretty far into the internals of Django for some operations. This is a large bit of work to port.
I’m happy to say that, after spending the day on it, I got surprisingly far toward having Django Evolution working on Django 1.8! It’s not perfect, and there’s not yet a good story for playing nice with Django migrations, but it’s a huge start. It opens the doors to getting a full compatibility story going.
The next question is, are we keeping Django Evolution, or moving to migrations fully? Well, that’s a bigger problem, because we have no control over which version of Review Board (and therefore Django) people are upgrading from, and have to be very careful with how we handle any database migrations.
There are also issues with Django’s migrations just being a lot slower than Django Evolution, to the intentional lack of an optimization step when applying the migrations. This means stupidly-long upgrades for large installs, which won’t work for us. So, we’ll probably stay with Django Evolution until we figure out a decent solution there…
