Looking at the branch:

One thing that I noticed is the extreme larger amount of, for passing the tests.

Branch_of_PR:
Executed 156 tests, with 0 failures (0 unexpected) in 52.261 (52.285) seconds

Master_branch (aerogear/aerogear-ios):
Executed 157 tests, with 0 failures (0 unexpected) in 12.194 (12.217) seconds





Now I am wondering, that that OHHTTPStub is really _that_ slow...
Do you have an numbers from the proposal #1 ?

-Matthias




On Tue, Jun 11, 2013 at 8:50 AM, Christos Vasilakis <cvasilak@gmail.com> wrote:
Hi

looks like #2 approach wins so I merged it.

Thanks!

On Jun 10, 2013, at 12:29 PM, Corinne Krych <corinnekrych@gmail.com> wrote:

+1 for #2. Indeed second approach2 is more objective-c in the syntax.



On 10 June 2013 10:15, Matthias Wessendorf <matzew@apache.org> wrote:
Hi,

I think I do prefer the approach #2 (the "mock helper" class)

-M


On Fri, Jun 7, 2013 at 10:32 AM, Christos Vasilakis <cvasilak@gmail.com> wrote:
Hi team,

for further improvements of our unit tests we have switched the http mocking mechanism we use (our own NSURLProtocol impl)  to the popular OHHTTPStubs[1] project,  a library currently recommended by the AFNetworking networking lib we use.

The basic mechanism is straightforward to use and encapsulated in one method:

return [OHHTTPStubsResponse responseWithData:data
statusCode:status
responseTime:responseTime
headers:headers];

in which a stubbed response is returned to the client.

Now, based on this mechanism,  we have abstracted a bit and created methods such as:

+ (void)mockResponse:(NSData*)data;
 + (void)mockResponseStatus:(int)status;
 + (void)mockResponseTimeout:(NSData*)data status:(int)status responseTime:(NSTimeInterval)responseTime;

This gives the advantages that a) clearly indicate what http scenario we are testing  and b) remove params that don't make sense for the particular scenario under testing e.g. that is we simulate a status of  (404) but we need to pass all params eg. data, interval, timeout, etc.  But this doesn't limit us, we can do that if we want and use the full blown method with all the params attached.

I have created two branches in my fork, one that uses a blocks approach inside the testing class [2] and one that the functionality is extracted in a helper class that the testing classes can use [3].  The second approach was created because there was common code and didn't want to duplicate it over the testing classes.  

I would be interesting to know what is your comments on it?

Thanks,
Christos


[4] 
 


_______________________________________________
aerogear-dev mailing list
aerogear-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev



--
Matthias Wessendorf

blog: http://matthiaswessendorf.wordpress.com/
sessions: http://www.slideshare.net/mwessendorf
twitter: http://twitter.com/mwessendorf

_______________________________________________
aerogear-dev mailing list
aerogear-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev

_______________________________________________
aerogear-dev mailing list
aerogear-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev


_______________________________________________
aerogear-dev mailing list
aerogear-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev



--
Matthias Wessendorf

blog: http://matthiaswessendorf.wordpress.com/
sessions: http://www.slideshare.net/mwessendorf
twitter: http://twitter.com/mwessendorf