Fix compatibility with Sanic 20.3.0

[brechtm]

This change breaks Sanic-CookieSession: sanic-org/sanic#1801

[danielalvsaaker]

I was about to open a proper pull request, but realized midway that it's not worth it considering the repository is not maintained. I would suggest migrating to sanic_session using the in-memory session interface.

[pyx]

Sorry about the hiatus.
@danielalvsaaker is you are still interested in using this, could you please reopen the pull request, I will make a new release based on that. Either way, thank you.

[brechtm]

Sorry about the hiatus.

No need to apologize! But a new release with this fix would be highly appreciated.

@danielalvsaaker is you are still interested in using this, could you please reopen the pull request, I will make a new release based on that. Either way, thank you.

As the owner of this project, you should be able to reopen the PR yourself.

[pyx]

@brechtm because PR #5 is broken: hard-coded session object name and inconsistency of object property vs dict element, and the new context type, types.SimpleNamespace does not have dict-like interface, etc. so I actually wrote a fix myself, including updated documentation, code example, and tests.

Then I found out the test client switch from requests to httpx, and that breaks the tests, e,g req, resp = app.test_client.get('/inc', cookies=cookies) does not keep the cookies in the object resp, half of the time (heisenbug😂), so I cannot simulate a series of post with states, httpx support context manager style code blocks (with statement) and it should work, except sanic 's test client does not support this nor expose the underlying httpx object.
Before I shoehorn an httpx object into tests, I need some reading about the difference between requests and httpx to find out why it doesn't work, I read the code in sanic briefly and it seems okay, passing the cookies object to httpx, I am feeling like debugging sanic and httpx at the same time...
I might rewrite the whole test-suit.

[pyx]

Now that I pushed the fix into master, close this one.