The argument that adding LiveServerTestCase with HTTPS as a separate tool is easy - it is valid in this case, after my refactor has been merged.
The real use-case when I (and someone else may) need HTTPSLiveServerTestCase is the one I wrote before:
- in a HTTPS-only application you can use custom secure cookies (not only CSRF or session, but application specific). But if you don't have a HTTPSLiveServerTestCase you cannot test it with Selenium as those cookies won't be passed to the server, which runs as HTTP
I'm not aware of any other issues, but maybe there is something more.
But my main point and what I wanted to suggest is that Django should encourage writing HTTPS-only applications. In my opinion HTTPS as default (instead of HTTP as default) should be seriously considered and thus, testing a HTTPS application shouldn't require to install external packages.
HTTPS is the future considering the SPDY/HTTP2 protocol development. Getting and deploying a certificate will also be much easier when Let's Encrypt launches in about a month from now.
This argument holds for runserver as well, but it's indeed another thread, so I will stick to running tests here :)
So "the problem" solved by HTTPSLiveServerTestCase is rather minor and can be achieved by using an external app.
But "the idea" to encourage HTTPS is another thing and adding a test case is a small step into this direction. And of course adding a HTTPS runserver would be a bigger step :)
Considering the above and the discussion about HTTPS runserver - I will just release the testcase as a separate app.