Re: [Webtest] JWebUnit style plugin - Ed Smith - com.canoo.lists.webtest

7 messages in com.canoo.lists.webtestRe: [Webtest] JWebUnit style plugin

From	Sent On	Attachments
Ed Smith	05 Jun 2007 11:50
Marc Guillemot	06 Jun 2007 04:19
Paul King	06 Jun 2007 04:46
Ed Smith	06 Jun 2007 05:31
Ed Smith	06 Jun 2007 05:39
Marc Guillemot	06 Jun 2007 06:04
Ed Smith	06 Jun 2007 07:18

Subject:	Re: [Webtest] JWebUnit style plugin
From:	Ed Smith (edsm...@yahoo.com)
Date:	06/06/2007 05:39:51 AM
List:	com.canoo.lists.webtest

You could factor out an HTMLUnit-independent interface for test engines that
included every feature of HTMLUnit, implement the plugin concept, and see if
anybody writes a plugin for any other engine (if they could make it fit the
interface). This should mitigate concerns over degrading to LCD. Maintain the
Selenium (or whatever) plugin as a subproject. If it falls behind, then it will
encourage its users to make a contribution without impacting the core WebTest.
Sort of a "Build it and see if they come" approach.

----- Original Message ---- From: Paul King <pau...@asert.com.au> To: webt...@lists.canoo.com Sent: Wednesday, June 6, 2007 6:46:44 AM Subject: Re: [Webtest] JWebUnit style plugin

Ed Smith wrote:

JWebUnit 2.X proposes the interesting idea of plugins. You specify tests in
JWebUnit lingo, but how the test is performed depends on the plugin. You can
choose from HTMLUnit or Selenium as your "real" tester. Of course, this is all
theoretical in JWebUnit as version 2.X isn't even working. Has any consideration
been given to taking this approach with WebTest?

WebTest was originally based on HttpUnit before we moved to HtmlUnit. At the time, we briefly considered whether we could make a plugin API so that you could plug in either of HttpUnit or HtmlUnit into WebTest. We decided against that as it tends to lead to (though it doesn't always have to) lowest common denominator APIs. By tightly integrating HtmlUnit we can make use of any features within it that we need. Also, over time, having multiple pluggable web engines requires considerably more time investment if you want to try to keep them all up to date. Maybe if we had 2 or 3 times as many committers we might be able to pull that off but it wouldn't be easy.