There seems to be concensus on the need for testing, and various
opinions for how to do it. As an example, I'll describe how the
WebLogicSubTask can be tested. Anybody's comments are welcome:
-Add the test code here:
xdoclet/core/test/xdoclet/ejb/vendor/weblogic/
|
+-build.xml
| +-ejbdoclet (test target)
| +-wlsdeploy (test target)
|
+-src/xdoclet/test/
|
+-CarEJB.java
+-WheelEJB.java
+-DriverEJB.java
In brief: Each xdoclet task impl should have a separate test directory
structure with sample Java sources with @task:tags. Further, there
should be a build script with targets that invoke XDoclet and possibly
tool-specific validation. The test is successful if the component(s)
can be generated, packaged, deployed and accessed without errors.
As Ara pointed out, it's hard to write test cases that cover all
combinations, so the test writer will have to be "clever" and write
cases (In my case, EJBs) that cover as much as possible. (For example,
I'll try to cover all kinds of CMR relationships in my test EJBs).
By separating tests this way, people can run the tests if they have the
appropriate environment. -If they don't, well, they can't. -But there
will always be someone who can.
Cheers, Aslak.
_______________________________________________
Xdoclet-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/xdoclet-devel
