Isn't clear how to use the testing backend about django-anymail HOT 3 CLOSED

I take it back about my tests failing that was just me being wrong as the code should have been:

msg = self.generate_email()
msg.send()

status = msg.anymail_status

email_sent = self.create_sent_object(email_id=status.message_id)

I'll do a pull request about adding to the docs though. Is the override_settings line above the correct way or should I be doing something else?

from django-anymail.

Django does indeed substitute its own special email backend during testing, by default.

I had actually intended Anymail's TestBackend for testing Anymail itself, and hadn't been thinking about other apps trying to test their use of Anymail. (That's why the TestBackend isn't documented. And why it has that weird required TEST_SAMPLE_SETTING, which was part of testing setting overrides.)

But what you're trying to do makes perfect sense, and it would probably be good to clean up and document the TestBackend. You're using it correctly with @override_settings. The ANYMAIL_TEST_SAMPLE_SETTING has to be there (for now), but can be set to anything.

from django-anymail.

FYI, I just committed a change that would let you run your tests with Django's normal testing backend. The one change you'd need to make is to create an AnymailMessage, rather than a EmailMultiAlternatives, in your generate_email function.

With this update, the AnymailMessage now initializes anymail_status immediately, so you won't get an AttributeError on a missing anymail_status if Django's test backend processes the message instead of Anymail. The status.message_id will be None in this case, but hopefully your create_sent_object function is already handling a missing message_id (since that can also occur in production if, say, your ESP has an error).

(I'd still like to clean up and document Anymail's TestBackend, and will leave this issue open to cover that enhancement. But hopefully this change makes using the TestBackend less necessary in the common case.)

from django-anymail.