imo, `@property` methods on #django models are to be avoided – *especially* when they trigger SQL queries. They seem harmless and convenient, but they're a common gateway to poor performance, model bloat, and hard to undo once baked into areas that typically lack adequate test coverage, like templates.
As an alternative, I ask, can the `@property` method be replaced by a `QuerySet` annotation?
Isn’t that what @cached_property is for? I use it *a lot*.
https://docs.djangoproject.com/en/4.2/ref/utils/#django.utils.functional.cached_property
@RichardBrockie cached property is great for reducing overhead when repeating calls on an instance. The n-query problem when iterating over a queryset remains.
This of course is often solvable with prefetch and/or select related, but it’s not always obvious when accessing a property that there’s a need for this.
@RichardBrockie A classic example of what I consider @property misuse is something like this:
@property
def total_count(self):
return self.other_model.count()
obj.total_count looks innocuous – it feels like a field, but it's costly in an iteration.
A preferred & performant alternative would be: `MyModel.objects.annotate(total_count=Count('other_model'))`.
Marc Gibbons
Richard Brockie