Index/Motion Commands

Discussion of the technology underlying the eValid solution.

Index/Motion Commands

Postby techr9 » Mon Jun 29, 2009 10:30 pm

I've been playing around with your "index/motion" commands...pretty slick.

How do you fit them in between your record/play capabiilty -- which is
pretty good, by the way...hat's off to you for that -- and your entirely scripted solution in C++, what you call the EPI version?
techr9
 
Posts: 5
Joined: Mon Jun 29, 2009 10:12 pm

Re: Index/Motion Commands

Postby eValid » Sat Jul 04, 2009 6:47 am

techr9 wrote:I've been playing around with your "index/motion" commands...pretty slick.

How do you fit them in between your record/play capabiilty -- which is pretty good, by the way...hat's off to you for that -- and your entirely scripted solution in C++, what you call the EPI version?


Thanks for the compliment. Of course, we love to hear that.

The reason this capability came about was due to AJAX applications that customers were asking us to test. While most of the time an AJAX application records OK and plays back OK it also is true that AJAX applications are notoriously sensitive to a changing page.

As you may know, eValid does have "adaptive playback" that corrects for common kinds of change in a page, but the AJAX pages were chaning very much more drastically than the adaptive playback algorithms could handle.

So we needed a way to raise the abstraction level one notch from where it already is. Without the index/motion commands eValid's architecture is basically an automated browser that you program from the eValid commands. So if you want to go to a URL you say GotoLink "URL" and if you want to click on an image you say SubmitClick "location on the screen". And so forth.

The next level up is to remove the requirement for specificity of what URL is -- when you want to go to a link -- and just identify it for where it is on the page.

Now if the page changes so much that you can't express how to find it from the index/motion commands, well, you have another problem -- because there isn't any higher level of abstration you can introduce.

How this capability fits into the record/play paradigm is this: The most efficient way to arrive at a reliable test is to make a recording "from life" andvthen when something breaks, you fix it. So you let the internal-state-based recording engine do its best to give you the script, but you augment it to improve the reliability sufficiently.

So that's where it fits: record from life, do fixups based on page properties, you've minimized the costs and maximized the return.

The eValid Programmatic Interface -- the EPI -- is just the command language for the product brought out as C++ functions/methods.

Hope this answer helps. Let us know if not.

The eValid Team
eValid
 
Posts: 2390
Joined: Tue Jan 01, 2008 12:48 pm
Location: USA


Return to Technology

Design Downloaded from free phpBB templates | free website templates | Free Web Buttons