[portal-commits] JBoss Portal SVN: r8716 - modules/test/trunk/docs/user-guide/en/modules.

Author: julien(a)jboss.com Date: 2007-10-19 08:47:43 -0400 (Fri, 19 Oct 2007) New Revision: 8716 Modified: modules/test/trunk/docs/user-guide/en/modules/introduction.xml Log: more documentation Modified: modules/test/trunk/docs/user-guide/en/modules/introduction.xml =================================================================== --- modules/test/trunk/docs/user-guide/en/modules/introduction.xml 2007-10-19 12:28:58 UTC (rev 8715) +++ modules/test/trunk/docs/user-guide/en/modules/introduction.xml 2007-10-19 12:47:43 UTC (rev 8716) @@ -10,7 +10,23 @@ <title>Introduction</title> <sect1> <title>Motivation</title> - <para>Todo</para> + <para>The JBoss Portal project historically has used JUnit as test framework. With the evolution of the project + its testing became more and more complex and required the development of several utility classes interacting + with JUnit. Those utilities became more complex and at some point the integration with JUnit required to + introduce hacks in order to get the testsuite of the project to execute.</para> + <para>When we decided to migrate the codebase to Java 5, the TestNG framework had emerged for a couple of years + and I hoped it would be the silver bullet that would solve my problems. After its study, I realized + that it was below my expectations in terms of integration.</para> + <para>Most of the existing unit test framework assumes that we test only Java objects which leads to frameworks + that expose object related idioms in their interface. A large part of the portal tests cannot be considered + as just Java objects and it would have been very difficult to integrate those tests with the existing frameworks.</para> + <para>Of course before I started this effort I tried to gather the pros and the cons of such an effort. The major + cons is about to not reinvent the wheel or that I should not spend my time developping something that not only + already exists but will require maintenance!!! A few things convinced me that I was right. The fact that the current + wheels don't meet my expectations and I spent much more time in the past to integrate JUnit with JBoss Portal codebase + rather than developping JBoss Unit. The fact that after all it is just a test framework and does not require much + maintenance, indeed JUnit was claimed to be invented in a few hours!!! Probably also the fact that it is just + a good thing to invest time in a tool that will make me save more time in the future.</para> </sect1> <sect1> <title>Principles</title>