MiddlewareSyncAsyncTests crashes with asgiref 3.3.2.

[Chris Jerdonek]

I see the following test failing on a recent PR: middleware_exceptions.tests.MiddlewareSyncAsyncTests.test_async_process_template_response_returns_none_with_sync_client
​https://djangoci.com/job/pr-mariadb/database=mysql,label=mariadb,python=python3.9/10979/

It looks like it might be because of the 3.3.2 asgiref release, since that was released today:
​https://github.com/django/asgiref/
and the failure is unrelated to what the PR changed.

I also see the same failure on a different recent PR:
​https://djangoci.com/job/pull-requests-bionic/database=mysql,label=bionic-pr,python=python3.8/11081/

[Chris Jerdonek]

I flagged it in asgiref's tracker: ​https://github.com/django/asgiref/issues/250

[Mariusz Felisiak]

​PR

[GitHub <noreply@…>]

In 78fea27f:

Fixed #32614 -- Fixed MiddlewareSyncAsyncTests tests with asgiref 3.3.2+.

[Mariusz Felisiak <felisiak.mariusz@…>]

In eb7c0a70:

[3.2.x] Fixed #32614 -- Fixed MiddlewareSyncAsyncTests tests with asgiref 3.3.2+.

Backport of 78fea27f690028204c03c28d821cb0c0240a7398 from main

[Chris Jerdonek]

Are the committed changes still needed with the fix that was deployed to asgiref in 3.3.3? If not, would it be cleaner to revert those?

[Andrew Godwin]

I would encourage them to be kept as the middleware should also be marked as an async function, regardless or not of how good asgiref is now at determining if something is async or not.

[Mariusz Felisiak]

Replying to Chris Jerdonek:

Are the committed changes still needed with the fix that was deployed to asgiref in 3.3.3? If not, would it be cleaner to revert those?

Yes, this is a separate issue. Django on the main branch doesn't support Python < 3.8, so the fix in asgiref doesn't affect it.