bbqs wrote:What do you mean by structural/geometric testing?
For an eValid user, the usual point of view is the things that you see on the face of the eValid browser. In normal "object mode" use eValid knows all about the properties of each object that you see in the internal terms that it sees. This information is facts such as window number, frame path, element index, ID tag of the element, and so forth.
For 99% of the cases, this information is more than sufficient to assure a very reliable playback. And when the page changes there is an adaptive playback feature that automatically "does the right thing".
But web applications can be very, very subtle and if you are working some some of the very advanced types of web applications -- including those that are AJAX based -- then even the detailed object mode information that an eValid script needs to behave in a reliable and predictable way may not be enough.
Enter "structural testing," which we also refer to as index/motion operation.
In this abstract mode the tester, guided by a recording from life, will augment the script with commands the locate page-invariant features by finding the index where they are on the current page. Then, again completely independent of the actual layout of the page, eValid commands can move the point of focus up or down a fixed number of elements, a number that is usually fixed for any one application.
Think of it as a way to point at "this thing on the page" no matter where that thing has been put this time around.
Now, you have your "pointer" at the right spot, and eValid can act on the object under your pointer; you can actually issue a full range (the complete possible range) of signals and events into the object you've pointed to.
A test done this way is very durable, very accurate, very reliable, and the extra computing needed to find and act on an event is less than 0.1% of the total system time eValid needs. So this mode is efficient, as well.
eValid Support Crew
by bbqs » Wed May 12, 2010 2:24 pm