Embedded Files
Challenges
Challenges
Navitation Model
Navitation Model
In general, navigation models are manually fine-tuned with human knowledge, which could be costly and impractical for testing large scale industray applications.
Example 1: Non-standard Input Generation
Example 1: Non-standard Input Generation
In Splittypie, the “transaction” page can only be discovered after typing in a time value with a specific format (mm/dd/yyyy), which is different from the standard W3C format. Thus, specific class definition (line 2-8) are leveraged in the navigation models, so that model-based algorithms thus can benefit from it when filling values (line 17).
Example 2: Specific Action Sequence
Example 2: Specific Action Sequence
Certain actions are combined as a micro action, which is executed in a specific order. Such a micro action has strong human knowledge, which is adopted in constructing the navigation models (line 9-18). This may be impractical for large scale industrial web applications, due to tremendous efforts in refining navigation models manually.
Concolusion:
Manual designing test cases is labor-intensive and costly, where the testing effectiveness heavily depends on the human testers’ domain knowledge.
Random-based approaches often create invalid test cases (like inputing on the button). Also, the testing is unbalanced and some hard-to-reach web pages may never be explored.
Model-based approaches requires non-trivial efforts in constructing the navigation model. Besides difference between the models and real applications restricting the testing effectiveness.
From a general perspective, test cases with long sequences of actions are hard for andom or search-based strategies to generate, thus an more intelligent testing algorithm is needed.
Page updated
Google Sites
Report abuse