I guess it's *capable* of doing all the stuff you mention.
But yeah, you'll need to script for it.
Only cross -browser tests I have run on Chrome & IE and, as you say, it simply runs the same set of tests twice. Once on Chrome, once on IE. I get sets of results back for each. I don't specifically highlight differences between the two But I could easily extend my framework to do so if need be. For our use, IE is a "nice to have" seconday consideration. We don't always run that part. Chrome is the key part.
The other issue I ran into with this is while the object model is almost always the same (I did run into a site built with Ruby where the DOM showed differences in TC's object browser - but it was work-round-able), some of the other elements of browsers/pages are not. How style information is reported differs between IE & Chrome. How you start the browser with a certain page is different. How you tell if page elements have fully loaded is different.
So yeah, the bulk of things worked. But I did have to code in some branching to account for areas where it didn't.
Mind you, browsers and sites are a lot more complex these days. Anything that is truly going to be 100% cross browser compatible, out of the box, would have to be very clever, and very complex! And would likely require frequent maintenace patches to keep it in line ....
Colin_McCrae wrote:I guess it's *capable* of doing all the stuff you mention.
...
y going to be 100% cross browser compatible, out of the box, would have to be very clever, and very complex! And would likely require frequent maintenace patches to keep it in line ....
Almost the same here, but I just started it. The test runs on IE and Chrome without having to change anything but passing the browser as parameter. The only problem are the screenshots...
colin: do you use screenshots?
If yes, how do you do this?
My solution (different threat hier in the forum) works ok, but is clearly not perfect.
If you build your test correctly, most probably it can run on more than one browser.
But you sure need scripting to setting up each test and stuff.
I only use screen shots to capture whats going on at the time a script encounters errors. And even then it's kind of a debug option which I can switch on and off in the project properties as needed.
Only image compares I do are ones where I compare an image element from screen against a stored control copy. And it's always a specific image, never a full screenshot. I've always found screenshot comparison tests to be way too prone to false positives as it's so easy for them to fail. Even with a mask.
There are certain CSS instruction thats the browsers render differently. For instance, a "fade" style in a panel header was clearly visible in a panel header bar as three distinct bands of colour in IE, but was a nice gradual fade - as it's meant to be - in Chrome. Which would have caused just about any screenshot compare in the site I test to fail.
As I say, basic elements and things are generally fine cross-browser. If you start inspecting the presentation (CSS - document stylesheet) layer though, thats where you're most likey to start hitting differences.
