Client mode return value

[asottile]

In client mode, the HTTPReporter was always returning None for its return value on the report method leading to testify in the client mode always doing exit(1). This changes the reporter to invariantly return True.

[bukzor]

I'd be more convinced if we improved the subprocess acceptance tests to cover this case; it's not clear to me how that return value relates to the exit code.

This is probably Good Enough tho. milki?

[ymilki]

I'm on the same page as @bukzor as to the relation between this return value and the exit code. Since you are changing None to True, are you alway implying that client mode will always return a True value (or at least not exit(1))?

What behaviours are we trying to change that relies on this? Should we include a note in a changelog (which we don't see to have o.O) or readme about this change in behaviour?

[asottile]

In the current case it returns 0 if the normal test reporter and the http reporter return True.

The normal test reporter returns True if all the tests pass and there are non-zero tests.

I'm not sure the http reporter should have a say in return value as it doesn't actually know or care whether tests pass or fail, and is mainly used as an interface shim to get at test results.

[bukzor]

This fix is fine, but I'd want to improve the tests such that it can be fixed in a more reasonable way in the future, ie extend dnephin's acceptance tests to also cover client+server.

[asottile]

Done. Tests are updated to cover both success and failure for client-server.

[bukzor]

Looks good to me. It fails when you break it yes?

[asottile]

of course

[bukzor]

Waiting for milki to get a say.

[ymilki]

lgtm @asottile. Thanks for adding the tests!