Observed on rev 183

Given a couple of models like the following

class Person(models.Model): id = models.AutoField(primary_key=True) name = models.CharField(max_length=50)

class PersonShoes(models.Model): id = models.AutoField(primary_key=True) person = models.ForeignKey(Person, related_name='shoes') brand = models.CharField(max_length=50)

If you wanted to query a list of people ordered by the number of shoes they own the django methodology for this would be:

peopleByShoeCount = Person.objects.annotate(num_shoes=Count('shoes')).order_by('num_shoes')

But when attempting this query with a django-pyodbc backend the following error is given:

[42S22] [Microsoft][ODBC SQL Server Driver][SQL Server]Invalid column name 'num_shoes'.

Quick correction. The error seems to only occur when attempting to slice the result set, like when attempting to get the top 10 results.

Person.objects.annotate(num_shoes=Count('shoes')).order_by('num_shoes')[:10]

Tojiro,

The problem is due to the fact that MS SQL doesn't allow you to use column aliases in the ORDER BY clause which is needed when you want to take a slice. Apply the attached patch to sql_server/pyodbc/query.py

This is an ugly patch but it works.