Best Practices for Testing iOS Mobile Applications: Go Cloud, Go Hybrid

iOS changed the mobility game, no doubt about it. It paved the way for the 'mobile era' by offering amazing functionality with a simple user experience.  However when it comes to testing and monitoring, working with the iPhone/iPad mobile application can be anything but simple…

As the iOS app market continues to produce record growth, challenges and complexities surrounding iOS application testing also continue to interfere with development. A key challenge of iOS testing is that, unlike the open-source Android OS, Apple iOS is a closed operating system. Added complexity during the development and testing stages arises with a closed system, since users can't extract necessary data from low level objects, which are essential for test automation. So, what's the best approach for getting the necessary level of access to the iOS device – rooting (jailbreaking) or compile-time source instrumentation? Should you base your testing on native objects or OCR-based screen analysis?

Let's take a deeper look into some of these challenges and why a cloud-based hybrid approach is important to offer developers and testers the necessary coverage, capabilities and flexibility to deliver better iOS apps and deploy them with confidence.

Rooting (jailbreaking) vs. Source Instrumentation (compile-time)

There are two common methods used today in the mobile testing industry to address this challenge (i.e. access to the low level objects): rooting (jailbreaking) and source instrumentation (i.e. compile-time solution).

Jailbreaking refers to the process of removing the limitations placed by Apple on the iOS device in order to get low level (root) access to the operating system. This allows the tester to be able to recognize the objects within the application being tested.

Source Instrumentation is performed by compiling the application being tested with an additional piece of code that provides access ("back door") to the low level OS for object recognition. This code enables the tester to execute the low level calls and get the Object ID's from the operating systems (without the need to root/jailbreak the device).

The decision as to what approach one should adopt strongly depends on several considerations (below are just few):

1)    The used SDLC process

2)    Corporate policies

3)    Application under test

4)    Frequency of testing

Perfecto Mobile provides its end users with the freedom to choose what fits them best, while taking into consideration the advantages and disadvantages of each approach. When customers need to quickly test either a new iOS version or a new iOS device, the jailbreaking approach is less suitable. In such a case, the compile-time method is preferable – even though it complicates the SDLC by introducing additional code to the application being tested.

On the other hand, using a jailbroken device lets you test the application with the exact code by which it will be released (compile-time mandates that before store submission, you will remove the "back-door" or be exposed to serious security issues). This eliminates the need for compilation and intrusive operations which could potentially pose a risk to quality. Companies using a compile-time approach should also consider possible regulations (such as HIPAA) which enforce testing on the final binary (and not on debug version, test friendly version, etc.)

The combined (hybrid) approach lets you choose which type of tests to implement on which iOS device according to the nature of your application, project needs, and policy. When the test devices are deployed and securely managed in a "private cloud" (such as that offered by Perfecto Mobile), such a configuration guarantees that the jailbreak method does not introduce any risks or abuse of the platform for non-testing purposes. The jailbroken device is used only for testing purposes in a closed and secure testing environment. This is analogous to the use the way iOS devices used for development have a "developer signature," as well as the way Android devices used for development have more levels of access than those required during the normal ALM cycle.

The Need for a Hybrid Approach to Object Recognition

Testing a mobile application requires strong object recognition capabilities. The use of visual analysis might not be sufficient, for example, the OCR technology can detect UI issues and glitches on the test devices, but cannot ensure 100% accuracy due to its heuristic nature. On the other hand, low level objects might "miss" the obvious qualification that a visual analysis can easily detect. That's why a hybrid approach incorporating both visual and Native object analysis is imperative for covering all mobile business cases. Such an approach is supported by Perfecto Mobile.

 

 

Object level analysis vs. Visual analysis

 

This screenshot above shows the differences of using an object level analysis as opposed to visual analysis (object level analysis would not have detected the overlapping of the button on the text).

The Perfecto Mobile Approach: Go Cloud, Go Hybrid

Perfecto Mobile's experience as a market leader has taught us that the best approach is to present each customer with all possible alternatives making them available inside the cloud.

Real devices + emulators (in the cloud),  OCR screen analysis + OS level native objects (in the cloud), rooted/jailbroken device + non-rooted/jailbroken devices (in the cloud)

With hundreds of thousands of automation hours running every month on our platform, we are well-positioned to suggest and guide, but not to "judge" what's best for everyone…

 

Perfecto Mobile hybrid object support on a rooted android and a non-jailbroken iPhone.