New post: Django Q objects with py3 and pytest

[jamescooke]

Fixes #6

Adds new post comparing Q objects in Python 3 and links the two posts.

[jamescooke]

@yanneves Can I get a merge?

[jamescooke]

@yanneves Can I get a merge?

[yanneves]

Nice! .rst is a new syntax to me. Looks good to merge.

[yanneves]

Nice! .rst is a new syntax to me. Looks good to merge.

[adamchainz]

assert isinstance(left, Q) and isinstance(right, Q) ??

https://stackoverflow.com/questions/1549801/differences-between-isinstance-and-type-in-python

[jamescooke]

I did look at using isinstance and made the decision purely on the output that pytest gives:

================================= FAILURES =================================
________________________________ test_fails ________________________________
test_assert_q_equal.py:112: in test_fails
    assert isinstance(a, Q)
E   AssertionError: assert False
E    +  where False = isinstance('Q', Q)
_______________________________ test_fails2 ________________________________
test_assert_q_equal.py:116: in test_fails2
    assert type(a) is Q
E   AssertionError: assert <class 'str'> is Q
E    +  where <class 'str'> = type('Q')
==================== 2 failed, 8 passed in 0.07 seconds ====================

Using type means that pytest explains what the type was that was found in the output and I found this more helpful.

Do you think I should abandon that idea and go for isinstance for the flexibility?

[adamchainz]

well you could just assert isinstance(a, Q), f'{a.__class__} is not subclass of Q'

[jamescooke]

That's a much better idea than either of mine! I'll try it out and PR it in.

[adamchainz]

you know, django does accept PR's

[jamescooke]

My guess would be that do to this well could take quite a large amount of effort while the payoff would be small. In the original post I wrote about why I thought this would not be implemented in Django: http://jamescooke.info/comparing-django-q-objects.html#the-perfect-world-predicate-logic

What do you think?

[adamchainz]

You could always argue that cmp could be done with str(self) == str(other) as a basic workaround like you have here

[jamescooke]

That sounds like an idea - I might give it a go 👍

[jamescooke]

I went and added my comment as a review like a GitHub n00b.

[jamescooke]

My guess would be that do to this well could take quite a large amount of effort while the payoff would be small. In the original post I wrote about why I thought this would not be implemented in Django: http://jamescooke.info/comparing-django-q-objects.html#the-perfect-world-predicate-logic

What do you think?

[jamescooke]

@yanneves Thanks for reading.

@adamchainz Thanks for reviewing.

I'm going to merge and publish but I'll update the post based on your comments Adam 😄

[jamescooke]

@yanneves Thanks for reading.

@adamchainz Thanks for reviewing.

I'm going to merge and publish but I'll update the post based on your comments Adam 😄