Django-了解RelatedManager删除方法

Question

So here I will be using classic Django Blog and Entry models from the documentation ( link ). I added null=True to the Entry's blog attribute.

>>> cb = Blog.objects.get(name__startswith="Cheese")
>>> gouda = Entry.objects.get(headline__startswith="Gouda")
>>> cb.entry_set.all()
[<Entry: Gouda>, <Entry: Emmentaler>]
>>> cb.entry_set.remove(gouda)
>>> gouda.blog
<Blog: Cheese blog>

I know that everything is fine and updated in database and if I query the second line from my example again that gouda.blog is gonna return None , but my question is why is gouda.blog not None without another query?

EDIT:

So if I understand everything correctly, this is how it works:

>>> cb = Blog.objects.get(name__startswith="Cheese")
>>> gouda = Entry.objects.get(headline__startswith="Gouda")
>>> cb.entry_set.all()
[<Entry: Gouda>, <Entry: Emmentaler>]
>>> cb.entry_set.remove(gouda)
>>> gouda.blog
<Blog: Cheese blog>

So default value of the bulk argument for the remove() method is True . That means that QuerySet.update() will be used. Object gouda will not be altered at the Python level so the blog attribute will still hold primary key of the "Cheese blog" blog. When we query gouda.blog , we will still get the "Cheese blog" object.

But what happens when bulk=False is passed to remove() ? From documentation: If bulk=False, the save() method of each individual model instance is called instead.

So then I override save() method of the Entry model like this:

def save(self, *args, **kwargs):
    super().save(*args, **kwargs)
    if self.blog == None:
        print(id(self))

and then:

>>> cb = Blog.objects.get(name__startswith="Cheese")
>>> gouda = Entry.objects.get(headline__startswith="Gouda")
>>> cb.entry_set.all()
[<Entry: Gouda>, <Entry: Emmentaler>]
>>> id(gouda)
139797518743592
>>> cb.entry_set.remove(gouda, bulk=False)
139797518745552
>>> gouda.blog
<Blog: Cheese blog>

Now we see that gouda object that save() method is called on is not the same as the one in our shell, so our object in the shell still holds primary key of "Cheese blog" in its blog attribute. And when we query for the blog with gouda.blog we still get "Cheese blog" object.

Is that correct and if it is, why is save() not called on the same object that we passed to remove() ?

Answer 1

You've said it yourself, you need to go back to the database to get the new information. The gouda object doesn't automatically keep a link to its database row; it only queries it when told to do so.