A Small-image Synchronization Method

How to use eValid to support regression testing.

A Small-image Synchronization Method

Postby newy » Wed Jan 12, 2011 4:11 pm

Why do you only provide a small-image synchronization method and not one that does a full screen?
newy
 
Posts: 1
Joined: Wed Jan 12, 2011 4:04 pm

Re: A Small-image Synchronization Method

Postby eValid » Mon Jan 17, 2011 11:03 am

newy wrote:Why do you only provide a small-image synchronization method and not one that does a full screen?

We did a lot of experimentation with the relative utility of different kinds of image synchronziation, and what we found out was that in practice one only RARELY ever needs to validate that an entire screen is invariant between regression test runs.

Instead, we found that if you start with a full screen you find you need to mask out areas of no concern...and that process was very complicated and difficult to implement.

So, we switch the point of view from "what has changed" to "what must be the same" and at the same time switch from a the outside in perspective to the inside out perspective. In other words, we implemented a "this must be the same" type of logic, using only a small fraction of the screen. Whence the "validate and synchronize on screen rectangle" command that eValid has.

Of course, if you HAVE to have the whole screen, you can expand the search region to as large an area as you want.

The eValid Support Team
eValid
 
Posts: 2392
Joined: Tue Jan 01, 2008 12:48 pm
Location: USA


Return to Regression Testing

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