Convert homespun tests to conventional tests #780
Conversation
akx
force-pushed
the
convert-strange-tests
branch
3 times, most recently
from
e889845
to
e7b0a34
Compare
|
pytest, not py.test since version 3.0 |
akx
changed the title
akx
force-pushed
the
convert-strange-tests
branch
3 times, most recently
from
5885aba
to
bf83885
Compare
old naming dies hard 😂 |
akx
force-pushed
the
convert-strange-tests
branch
from
bf83885
to
d54cea3
Compare
|
Ah, I'll improve this a little more – we'll want to be sure the client process didn't die with an exception and so on... Also turns out the original author(s) of the clients have been funny with where they actually mean |
akx
force-pushed
the
convert-strange-tests
branch
5 times, most recently
from
63f1758
to
e381f43
Compare
akx
force-pushed
the
convert-strange-tests
branch
from
e381f43
to
7720250
Compare
akx
force-pushed
the
convert-strange-tests
branch
from
7720250
to
2413572
Compare
|
Phew, this PR underwent a bit of a metamorphosis since last night, but I think it's much, much better now. We will now be able to automagically catch exceptions from the clients, and there's (almost) no playing with return codes and so on. I also fixed the way port A bunch of actually unused/no-op code also got removed – a 723 line removal is pretty good in my books 😁 I think there's still more unused code, but the MQTTv5 tests would need to be fixed first (#769) to see if that's the case; they don't actually run against the current |
| @@ -25,13 +24,26 @@ class TestError(Exception): | |||
| def __init__(self, message="Mismatched packets"): | |||
| self.message = message | |||
|
|
|||
|
|
|||
| def bind_to_any_free_port(sock) -> int: | |||
There was a problem hiding this comment.
❤️ I will no longer have strange error when I run a real broker on port 1888 and run paho test.
| PYTHONPATH=f"{tests_path}{os.pathsep}{os.environ.get('PYTHONPATH', '')}", | ||
| ) | ||
| assert 'PAHO_SERVER_PORT' in env, "PAHO_SERVER_PORT must be set in the environment when starting a client" | ||
| # TODO: it would be nice to run this under `coverage` too! |
There was a problem hiding this comment.
We might be able to have coverage done it instead of a subprocess we start a thread. But this is probably easy to said and harder to implement. Anyway I think it not in the scope of this PR.
There was a problem hiding this comment.
pytest-cov actually hooks things up so subprocesses work OOTB, which was nice to notice!
No problem, it was a nice brain workout while watching Christmas movies and drinking mulled wine and cognac... 😂 |
This PR converts the homespun
test/code to (what I think is) pretty idiomatic Python/Pytest code.I think I got everything right, the original code was pretty hairy. 😂
Would probably be even better if the client code was included in the tests themselves to avoid having to hop between files.
Fixes #768