selenium-tests-other use a testing framework put together by Red Hat's EPP QE team. Using an API directly gives a better control to a test writer, and the tests themselves can be finetuned with wait times and such to be more robust in the face of occasional congestion ... so they don't needlessly fail due to occasional long response times, producing a false negative on the test.

It's not very hard to use browser debug tools (i.e. Firebug, or integrated DOM inspector/console) to target specific elements by id, or a more logical parent-element-relative xpath queries that make the tests much more resilient to changes in HTML. 

One gripe with selenium-snifftests is also that they are organized by using tremendously long file names / paths, which causes problems for windows users. (This is not really due to the technology, but due to choice of naming and organizing tests).


On Fri, Jun 15, 2012 at 9:06 AM, Nguyen Anh Kien <kienna@exoplatform.com> wrote:
Hi,

Iam doing improvements for GateIn selenium. There are some tests in gatein-portal/testsuite/selenium-tests-other, they are writting in Java. Iam wondering what advantages between writing Java testcases directly AND using Selenium IDE to record testcases. 

In eXo, we are using Selenium IDE to record test scripts in html patterns, looking at gatein-portal/testsuite/selenium-snifftests, and then use selenium-maven-plugin to convert automatically htmls to Java test files.

What mechanism should we use to have quickly to write Selenium testcases, easily to maintain, etc? 

regards,
================
Kien Nguyen
M&E: (+84) 933.975.960 - kienna@exoplatform.com
Skype & yahoo: nakien2a
Portal Team - eXo Platform South East Asia (http://www.exoplatform.com)
A: 7 Flr, ThaiHa building, 18/11 alley, ThaiHa Str, Hanoi, Viet Nam
_______________________________________________
gatein-dev mailing list
gatein-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/gatein-dev