AliSoftware
commented
8 years ago Yes sadly it's a drawback from how NSURLProtocol is used in the Cocoa URL Loading system, calling the protocol's canInitWithRequest: of the protocol twice sometimes (depending on iOS version) for some reasons.
I have no control over it because that's a method called by the OS itself twice that triggers this double call so no real way to avoid this properly.
That's also a reason why I did a refactor the API some major versions ago — before the current API, the old one only took one block/closure and stubbed the request only if you returned non-nil response, but the fact that the methods was called many times by the OS made your potentially expensive code in that closure (maybe reading the content to sub from a file for example) run multiple times :-/ . That's another reason why those two closures (the one for testing and the one for building the response) are now separate.
I've filed a bug report to Apple years ago about that behavior but no response so far.
markedwards
Using both 5.1.0 and 5.2.0, when YES is returned from the test in stubRequestsPassingTest:withStubResponse:, I see a duplicate request passed to the test. If NO is returned, this does not occur. By duplicate, I mean the exact same request object is passed to the test a second time.
This seems to be due to two calls to the firstStubPassingTestForRequest: method in OHHTTPStubs.m. Is this the intended behavior? It seems like it should be unnecessary to run every test twice.