Hi,
Paul has mentioned the excellent reporting. I'd like to insist on this
point as I think that it is a killer argument. In the report you see
everything that has been done including answers received from server and
you can browse these answers. When everything works fine it's useless
but would it be always the case, then no automated test would be needed.
Understanding why a test doesn't work (anymore) is a report browsing
work with WebTest whereas it is a java debugging activity with JWebUnit.
This make a huge difference when for instance your test has 100 steps
and fails at step 90 due to a problem at step 80.
Marc.
Paul King wrote:
JWebUnit used to be a very thin veneer over the top of HttpUnit
but they have moved to HtmlUnit and the next version will also
support Selenium RC as an alternative plugin.
If all of the people who develop your tests are Java developers
and sit in front of an IDE all day, then it isn't a bad choice.
WebTest is possibly more accessible to a wider range of
testers but mostly it is known for:
+ its excellent reporting (jwebunit might give more immediate
feedback to Java developers used to the green bar but has nothing
equivalent to WebTests reports which can be used as acceptance
documentation presentable to customers)
+ its robust implementation (excellent test coverage)
+ its breadth of applicability (not only Web Pages but
also Web Services, PDF files, Excel files, Email systems etc.)
+ its extensibility through Ant's many available tasks
and through scripting languages
+ its breadth of capability for testing Web Applications, e.g.
its xpath and regex support etc. which don't appear to be as
strong in jWebUnit.
Given that you have existing tests written using WebTest, I would
have thought you would need some gains from jWebUnit over WebTest
to justify swapping.
If managing tests is an issue you should be consider refactoring
your tests to remove duplication. If you are using the XML flavour
of WebTest, this amounts to defining your macros. If you are using
the Groovy flavour of WebTest, this amounts to writing your own
higher level functions above the WebTest 'API'. jWebUnit and WebTest
in some sense provide higher-level refactorings above HtmlUnit but
they are not at the level you should be working at, you should be
thinking in terms of the domain of your application.
Regards, Paul.
Ejaz Ahmed Syed wrote:
A passionate colleague of mine thinks managing canoo tests is pain and
strongly believes that JWebunit is the way to go to test web
applications. (I know, !!) He wants to write a recorder similar to
TestGen4Web that can be used in firefox to record tests.
I was wondering if anyone has thoughts one vs other, may be some one
here might have looked at JWebunit at closer detail before they
started using Canoo and can tell how badly it sucks. it would have
been a good conversation if he was promoting HttpUnit, but thats not
the case. I know very little about JWebunit to argue other than that I
have lots of working integration-tests, and happy developers.
_______________________________________________
WebTest mailing list
[email protected]
http://lists.canoo.com/mailman/listinfo/webtest
_______________________________________________
WebTest mailing list
[email protected]
http://lists.canoo.com/mailman/listinfo/webtest
