Fair question, but also a very broad question.
Web applications that employ AJAX can range from very simple (for example, a simple autocomplete activity), to the very, very complex (for example, the recently introduced
http://www.outlook.com email system from Microsoft which they are pitching as better that Gmail).
At a very high level we have found this method of using eValid to test and measure AJAX applications to be the most effective:
(1) Make a recording from life first. "From life" means, using the out-of-the-box settings, make a recording from the starting URL to the point where you are going to be validating a result. The Wait times that eValid puts into your script will, at playback time, provide a crude level of synchronization.
(2) Play the recording back...it ought to work the first time assuming you have not edited the script and are not multiplying the Wait times by the Wait Time Multiplier to make things "go faster."
(3) Now, play back the script increasing the speed -- and increasing the changes for a script de-sync due to AJAX. When the script de-sync's, mark that spot in the playback, go back to that point, and record some kind of synchronization step. Repeat this until you can run your script with the Wait Time Multiplier = 0 -- that is, with NO wait times, but only synchronizations.
To a first level, this will get you a good script. Of course it can become more complex if you want to do structural testing on the page or if you want to replace some parts of your script with fancy DOM manipulations.
eValid Support
by KChance » Tue May 28, 2013 4:29 pm