I recently upgraded to 1.1 because we needed the new aggregation/annotation features. For instance, this is a query we were doing before:

    articles = Article.objects.filter(share__sharedOn__range=date_range).distinct()
    articles = sorted(articles,key=lambda k: k.TotalRecipients(date_range),reverse=True)

which was achingly slow because it involved making a couple thousand SQL queries during the sorting phase. With annotation we can just do:

    articles = Article.objects.filter(share__sharedOn__range=date_range)\
    .annotate(total_recipients=Count('shares__recipients')).order_by('total_recipients')
    

which is much much faster and more efficient. Without annotation, we would have had to manually write the SQL or denormalize a bunch of ManyToMany fields, neither of which I was too excited about.

I have a suspicion that you might have been able to do it more efficiently (e.g., in one query) using extra() to provide a similar effect to the annotation. That's how I typically handled this sort of thing prior to aggregates landing.

I don't know how you Python guys, with your drive for readability and elegance which I admire, can tolerate so many underscores in your OOP code. This looks like C preprocessor to me.

It makes sense a lot, since you can't have comparison in the parameters (you'll pass a boolean instead of the comparison).

I totally understand you, but, I think its the best way to go, of course this is one example, and its very complicated, most of the time you'll only need one of those parameters.

Release notes :

http://docs.djangoproject.com/en/dev/releases/1.1-beta-1/#re...

Proxy models, deferred models seem like awesome features.

"The dumpdata management command now accepts individual model names as arguments, allowing you to export the data just from particular models."

That'll be nice for testing.