dispatch_semaphore_t ignored about specta

[jspahrsummers]

A few things here:

1. You're creating the semaphore with an initial count of 1. This means that the first dispatch_semaphore_wait will always complete immediately, which is probably not what you want.
2. The semaphore isn't recreated each time the beforeAll block is run (relevant if it's in a describe block). I'd probably move the variable and initialization into there.
3. If -loginWithUserName:… needs to do any work on the main thread — like delivering its results there — it won't be able to, since you'll be blocking on the semaphore.

Instead, if you're also using expecta, you can use the will matcher to solve all of the above problems:

beforeAll(^{
    // This is run once and only once before all of the examples
    // in this group and before any beforeEach blocks.

    // Set a timeout of 100 sec
    setAsyncSpecTimeout(100);

    Login *login = [[Login alloc] init];

    __block BOOL completed = NO;
    [login loginWithUserName:@"user"
                    passWord:@"pass"
                   onSuccess:^(id responseObject) {
                       // Handle successful login
                       completed = YES;
                   }
                     onError:^(NSError *err) {
                       // Handle error
                         completed = YES;
                     }];

    expect(completed).will.beTruthy();
});

As a sidenote, 100 seconds is way too long to be waiting for results in a unit test. It's probably safe to say that any more than 10 seconds should be considered a failure.

[cyupa]

The timeout was left that way because of a very big lag in the first place. I changed that but it doesn't make any difference to the issue.

Yes, I am also using specta but it fails. You could try for yourself the example you gave and see if it fails, for me it does. I added some brakepoints in both success and failure callbacks, they are never called.

Ok, so the semaphore would block, I'll drop it.

Now, following your suggestion I'm having something like this:

// The authToken that will be used in all the tests
__block NSString *authToken = nil;

beforeAll(^{
        // This is run once and only once before all of the examples
        // in this group and before any beforeEach blocks.

        // Set a timeout of 20 sec
        setAsyncSpecTimeout(20);

        UserSession *sessionManager = [UserSession sharedInstance];
        [sessionManager userSessionWithUsername:@"user"
                                          password:@"pass"
                                          success:^(UserSessionData *session) {
                                                 // Set the authToken string
                                                 authToken = session.authToken;
                                     }
                                          failure:^(NSError *error) {
                                                 // Handle the error
                                    }];

        expect(authToken).will.beInstanceOf([NSString class]);

    });

The will matcher fails. It tries to do the other tests, which also fail, because they make use of the authToken.

[jspahrsummers]

I added some brakepoints in both success and failure callbacks, they are never called.

This sounds like a problem with the method you're using, then. Does it work in a small sample application?

[orta]

Closing due to lack of a response.