Fix doc test for async_client#262
Conversation
The doc test for the async_client::wait function assumed an MQTT broker to be running. This commit updates the test to assert that an error is returned when waiting on the connection being established, which should usually be the case.
|
this fix is also included in #260 |
|
I'm preparing a release to go out next week, so a bunch of things will break and be fixed in the next few days. For the CI, my hope was to spin up a version of Mosquitto in the Actions containers to test against. |
Good to know. I do have a fix for #244 that I would really like to get into that release, if you don't mind.
That's fine for running the tests on CI, but when running in your local workspace, it feels odd that executing the tests fails ... |
Yeah, sorry. You don't know what is in the branches I haven't pushed yet! :) For just this one doc test we can ignore the error. I started another branch (not pushed yet) to to implement and test some improved error handling. For that I want to have unit regression tests to make sure the error results stay consistent. But I don't plan on setting up mocks to do it; so I need to test against a broker. What I can do (and have done on some other crates) is to have a cargo build feature, like "broker-tests", which would require a compliant broker on localhost to pass. That way these tests would be opt-in, so shouldn't surprise anyone. And then set up the CI to run the broker in the container. |
|
Oh, just to clarify, I will close this and set the doc comment to ignore. I would rather not have code in the source comments that looks like it shouldn't pass. I worry that might be confusing. |
2 participants
The doc test for the async_client::wait function assumed an MQTT broker to be running.
This commit updates the test to assert that an error is returned when waiting on the connection being established, which should usually be the case.