Re-add execution of doctests #768
Conversation
This comment was marked as spam.
This comment was marked as spam.
mfussenegger
force-pushed
the
j/doctests
branch
from
0addb72 to
d6663b9
Compare
mfussenegger
force-pushed
the
j/doctests
branch
from
d6663b9 to
1a80b98
Compare
| respond, the request is automatically routed to the next server: | ||
|
|
||
| >>> invalid_host = 'http://not_responding_host:4200' | ||
| >>> invalid_host = 'http://127.0.0.1:4201' |
There was a problem hiding this comment.
Haven't investigated why, but on my machine resolving a hostname that doesn't exist takes way longer than having it resolve to localhost on a port where hopefully nothing runs.
Potential follow up would be to generally restructure this. http.rst and client.rst have lots of overlap.
| >>> invalid_host = "invalid_host:9999" | ||
| >>> even_more_invalid_host = "even_more_invalid_host:9999" | ||
| >>> http_client = HttpClient([crate_host, invalid_host, even_more_invalid_host], timeout=0.3) | ||
| >>> invalid_host1 = "192.0.2.1:9999" |
There was a problem hiding this comment.
| makeSuite = unittest.TestLoader().loadTestsFromTestCase | ||
|
|
||
|
|
||
| def test_suite(): |
There was a problem hiding this comment.
pytest afaik doesn't support the test_suite discovery. Tests are picked up based on filename pattern and LayerTest/LayerUtilsTest still get executed
tests/test_docs.py
Outdated
| print(s) # noqa: T201 | ||
|
|
||
|
|
||
| def test_docs(doctest_node): |
There was a problem hiding this comment.
I also tried --doctest-glob="*.rst" (https://docs.pytest.org/en/stable/how-to/doctest.html) instead with a autouse fixture, but that would also be used in the other tests and we'd have two crate instances running in parallel.
So this seemed like the better option.
Follow up to #767
mfussenegger
force-pushed
the
j/doctests
branch
from
1a80b98 to
29a1fa9
Compare
Follow up to #767