Taking advantage of WR in web testing ...
I've just finished scripting my first complete web application regression test which will be run after each change, and every 90 days even with no changes. This app is an ASP auction site and can reside on three different servers - DEV, STAGING, PRODUCTION - and I am using the Per Test GUI Map Mode. The app is broken down into seven smaller modules and here's how I am testing each module ...
* Login and manually nav to the HOME page
* Test the URL to populate the server variable, nav to the module under test, and begin GUI checkpoint verification (links) based on which server you're on
* Perform each functionality and test for success by verifying text (get_text) confirmation
* Then begin nav'ing to each subsection and test all the GUI checkpoints (based on which server) and functionality, etc.
* This continues for each module, then a results report is printed and a defect density calculated and the test packet is filed, etc.
Is this, generally speaking, how more experienced testers use WinRunner for web testing? What do you guys do differently? Is there an advantage to using database checks to ensure an operation was successful, as opposed to checking the text of success/completion messages?
For those who take the time to offer input ... THANKS ever so much! Your time is greatly appreciated.
Phil Minh Thong
Re: Taking advantage of WR in web testing ...
It appears that you have a good approach - especially if you feel you have tested adequately and you have included the introduction of errors, and - subsequent validation of error handling.
How experienced test engineers would handle your case may vary and therefore there is no precise answer. Do your tests satisfy the test objectives?
Using database checks can be a good method of verification. Bear in mind that the functions that support those are there to make it more convenient for you to implement verification expeditiously. Other methods of verification (such as going directly to the database via "db_*" functions) might be more practical depending on your test coverage objectives. Are there risks with not checking the database? Can you miss something by focusing only on front-end?
Other considerations (and there are many others) include regression suite maintenance objectives. It is desirable to have low-maintenance scripts.
You might want to browse the Web Testing forum for more specifics and references.
[This message has been edited by jpensyl (edited 09-13-2002).]