Django .values() after F() expression filter

Question

It seems that Django can not group values after you use a filter with an F() expression. For example:

(models.py)

class Athlete(models.Model):
    name = models.CharField(max_length=30)
    nationality = models.CharField(max_length=30)

class TrainingSession(models.Model):
    training_date = models.DateTimeField()
    location = models.CharField(max_length=30)
    athlete = models.ForeignKey(Athlete)

class Run(models.Model):
    run_time = models.IntegerField()
    training = models.ForeignKey(TrainingSession)

(views.py) I want the number of runners for every location where they had their best run.

r = Run.objects.all()

# Annotate the best time from the training session and filter
r = r.annotate(best_run_time = Min('trainingsession__run__run_time'))
r = r.filter(run_time = F('best_run_time'))

# Group by location
r = r.values('location')
r = r.annotate( athlete_count = Count('trainingsession_athlete', distinct=True))

When I don't use the F() expression, the .values('location') will group the result per location. However, when I use it, every location is given multiple time with an athlete_count of 1.

Answer 1

First of all, r.annotate(best_run_time = Min('trainingsession__run__run_time')) will annotate a different time to every Run object which is not what you expect. You want to include everything in the queryset, but annotating does it for each object separately, so if a Run object has a run_time of 10 units, then 10 is also annotated. What you really need is aggregate() that you can use like this:

><> best = Run.objects.all().aggregate(Min('run_time'))
><> best
>>> {'run_time__min': 10}

Your query could work like this:

Run.objects.filter(run_time=best['run_time__min']).values('location')  

There are some problems with Django querysets when using F , annotations and filtering, but without knowing which Django version you are using, it's kinda hard to point you in the correct direction.