we have a web/browser based GUI application.
we have a requirement that it should support
WINDOWS NT, 2000 for IE and netscape.
Also, Netscape browser on Linux platform
My question is:
Isnt it enough, if we do a full functional testing of the apllication for both IE and netscape on Windows?.
Do we also need to do testing on Linux platform too? (as it is a browser based application).
The reason i am asking is: we have winrunner software for performing automation of tests on windows side. And i think winrunner doesnt go with Linux.
Sadly enough, I've seen several bugs where it works under NC 4.7 in Windows, but for some reason doesn't in Linux (and vice-versa). If that's the requirement, then you should perform testing in all supported environments.
Obviously you don't want to have to run full regression five times. The way I handle this is to divide up my testing team so that some of them are testing on one browser/platform, and others on another. If you rotate the regression tasks around, then eventually all of the regression gets done on all platforms, and you are not completely eliminating testing on any given platform in a regression cycle.
Question "Isnt it enough, if we do a full functional testing of the apllication for both IE and netscape on Windows?."
Answer : 99 % yes since all other available browsers are mostly built on top of these two. And "most" of the users will be using only these two as internet browser
Question "Do we also need to do testing on Linux platform too?"
Answer : Yes since netscape in linux and windows sometime act diffrently
As far as my experience goes. Yes we need to do testing on both the platforms compulsory, because when you test the application on Windows NT / 2K. The same application if you run on Windows 98, it will give strage behaviour.
Why i am saying is this, because linux is all together a different O/S. You need to do the Testing on Linux too.
As you mentioned you already have WinRunner Scripts for Windows. So I guess you need to put your effort for Linux testing only.